Search

Current reinforcement-learning methods are unable to directly learn policies that solve the minimum cost reach-avoid problem to minimize cumulative costs subject to the constraints of reaching the goal and avoiding unsafe states, as the structure of this new optimization problem is incompatible with current methods. Instead, a surrogate problem is solved where all objectives are combined with a weighted sum. However, this surrogate objective results in suboptimal policies that do not directly minimize the cumulative cost. In this work, we propose RC-PPO, a reinforcement-learning-based method for solving the minimum-cost reach-avoid problem by using connections to Hamilton-Jacobi reachability. Empirical results demonstrate that RC-PPO learns policies with comparable goal-reaching rates to while achieving up to 57% lower cumulative costs compared to existing methods on a suite of minimum-cost reach-avoid benchmarks on the Mujoco simulator. The project page can be found at https://oswinso.xyz/rcppo., Comment: Accepted to NeurIPS 2024

While large language models (LLMs) have recently demonstrated strong potential in solving planning problems, there is a trade-off between flexibility and complexity. LLMs, as zero-shot planners themselves, are still not capable of directly generating valid plans for complex planning problems such as multi-constraint or long-horizon tasks. On the other hand, many frameworks aiming to solve complex planning problems often rely on task-specific preparatory efforts, such as task-specific in-context examples and pre-defined critics/verifiers, which limits their cross-task generalization capability. In this paper, we tackle these challenges by observing that the core of many planning problems lies in optimization problems: searching for the optimal solution (best plan) with goals subject to constraints (preconditions and effects of decisions). With LLMs' commonsense, reasoning, and programming capabilities, this opens up the possibilities of a universal LLM-based approach to planning problems. Inspired by this observation, we propose LLMFP, a general-purpose framework that leverages LLMs to capture key information from planning problems and formally formulate and solve them as optimization problems from scratch, with no task-specific examples needed. We apply LLMFP to 9 planning problems, ranging from multi-constraint decision making to multi-step planning problems, and demonstrate that LLMFP achieves on average 83.7% and 86.8% optimal rate across 9 tasks for GPT-4o and Claude 3.5 Sonnet, significantly outperforming the best baseline (direct planning with OpenAI o1-preview) with 37.6% and 40.7% improvements. We also validate components of LLMFP with ablation experiments and analyzed the underlying success and failure reasons., Comment: 50 pages, 25 figures, 7 tables

Control Barrier Functions (CBFs) have proven to be an effective tool for performing safe control synthesis for nonlinear systems. However, guaranteeing safety in the presence of disturbances and input constraints for high relative degree systems is a difficult problem. In this work, we propose the Robust Policy CBF (RPCBF), a practical method of constructing CBF approximations that is easy to implement and robust to disturbances via the estimation of a value function. We demonstrate the effectiveness of our method in simulation on a variety of high relative degree input-constrained systems. Finally, we demonstrate the benefits of RPCBF in compensating for model errors on a hardware quadcopter platform by treating the model errors as disturbances. The project page can be found at https://oswinso.xyz/rpcbf., Comment: Submitted to ICRA 2025. The project page can be found at https://oswinso.xyz/rpcbf

Prediction of failures in real-world robotic systems either requires accurate model information or extensive testing. Partial knowledge of the system model makes simulation-based failure prediction unreliable. Moreover, obtaining such demonstrations is expensive, and could potentially be risky for the robotic system to repeatedly fail during data collection. This work presents a novel three-step methodology for discovering failures that occur in the true system by using a combination of a limited number of demonstrations from the true system and the failure information processed through sampling-based testing of a model dynamical system. Given a limited budget $N$ of demonstrations from true system and a model dynamics (with potentially large modeling errors), the proposed methodology comprises of a) exhaustive simulations for discovering algorithmic failures using the model dynamics; b) design of initial $N_1$ demonstrations of the true system using Bayesian inference to learn a Gaussian process regression (GPR)-based failure predictor; and c) iterative $N - N_1$ demonstrations of the true system for updating the failure predictor. To illustrate the efficacy of the proposed methodology, we consider: a) the failure discovery for the task of pushing a T block to a fixed target region with UR3E collaborative robot arm using a diffusion policy; and b) the failure discovery for an F1-Tenth racing car tracking a given raceline under an LQR control policy., Comment: 8 pages, 7 figures

While a lot of recent research focuses on enhancing the textual reasoning capabilities of Large Language Models (LLMs) by optimizing the multi-agent framework or reasoning chains, several benchmark tasks can be solved with 100% success through direct coding, which is more scalable and avoids the computational overhead associated with textual iterating and searching. Textual reasoning has inherent limitations in solving tasks with challenges in math, logics, optimization, and searching, which is unlikely to be solved by simply scaling up the model and data size. The recently released OpenAI GPT Code Interpreter and multi-agent frameworks such as AutoGen have demonstrated remarkable proficiency of integrating code generation and execution to solve complex tasks using LLMs. However, based on our experiments on 7 existing popular methods for steering code/text generation in both single- and multi-turn settings with 14 tasks and 6 types of LLMs (including the new O1-preview), currently there is no optimal method to correctly steer LLMs to write code when needed. We discover some interesting patterns on when models use code vs. textual reasoning with the evolution to task complexity and model sizes, which even result in an astonishingly inverse scaling law. We also discover that results from LLM written code are not always better than using textual reasoning, even if the task could be solved through code. To mitigate the above issues, we propose three methods to better steer LLM code/text generation and achieve a notable improvement. The costs of token lengths and runtime are thoroughly discussed for all the methods. We believe the problem of steering LLM code/text generation is critical for future research and has much space for further improvement. Project Page, Datasets, and Codes are available at https://yongchao98.github.io/CodeSteer/., Comment: 32 pages, 12 figures, 12 tables

Navigating rigid body objects through crowded environments can be challenging, especially when narrow passages are presented. Existing sampling-based planners and optimization-based methods like mixed integer linear programming (MILP) formulations, suffer from limited scalability with respect to either the size of the workspace or the number of obstacles. In order to address the scalability issue, we propose a three-stage algorithm that first generates a graph of convex polytopes in the workspace free of collision, then poses a large set of small MILPs to generate viable paths between polytopes, and finally queries a pair of start and end configurations for a feasible path online. The graph of convex polytopes serves as a decomposition of the free workspace and the number of decision variables in each MILP is limited by restricting the subproblem within two or three free polytopes rather than the entire free region. Our simulation results demonstrate shorter online computation time compared to baseline methods and scales better with the size of the environment and tunnel width than sampling-based planners in both 2D and 3D environments., Comment: Accepted by IEEE RA-L. URL: https://sites.google.com/view/realm-rigidmilp

The mixed truck-drone delivery systems have attracted increasing attention for last-mile logistics, but real-world complexities demand a shift from single-agent, fully connected graph models to multi-agent systems operating on actual road networks. We introduce the multi-agent flying sidekick traveling salesman problem (MA-FSTSP) on road networks, extending the single truck-drone model to multiple trucks, each carrying multiple drones while considering full road networks for truck restrictions and flexible drone routes. We propose a mixed-integer linear programming model and an efficient three-phase heuristic algorithm for this NP-hard problem. Our approach decomposes MA-FSTSP into manageable subproblems of one truck with multiple drones. Then, it computes the routes for trucks without drones in subproblems, which are used in the final phase as heuristics to help optimize drone and truck routes simultaneously. Extensive numerical experiments on Manhattan and Boston road networks demonstrate our algorithm's superior effectiveness and efficiency, significantly outperforming both column generation and variable neighborhood search baselines in solution quality and computation time. Notably, our approach scales to more than 300 customers within a 5-minute time limit, showcasing its potential for large-scale, real-world logistics applications.

Verifying safety of neural network control systems that use images as input is a difficult problem because, from a given system state, there is no known way to mathematically model what images are possible in the real-world. We build on recent work that considers a surrogate verification approach, training a conditional generative adversarial network (cGAN) as an image generator in place of the real world. This enables set-based formal analysis of the closed-loop system, providing analysis beyond simulation and testing. While existing work is effective on small examples, excessive overapproximation both within a single control period and across multiple control periods limits its scalability. We propose approaches to overcome these two sources of error. First, we overcome one-step error by composing the system's dynamics along with the cGAN and neural network controller, without losing the dependencies between input states and the control outputs as in the monotonic analysis of the system dynamics. Second, we reduce multi-step error by repeating the single-step composition, essentially unrolling multiple steps of the control loop into a large neural network. We then leverage existing network verification tools to compute accurate reachable sets for multiple steps, avoiding the accumulation of abstraction error at each step. We demonstrate the effectiveness of our approach in terms of both accuracy and scalability using two case studies: an autonomous aircraft taxiing system and an advanced emergency braking system. On the aircraft taxiing system, the converged reachable set is 175% larger using the prior baseline method compared with our proposed approach. On the emergency braking system, with 24x the number of image output variables from the cGAN, the baseline method fails to prove any states are safe, whereas our improvements enable set-based safety analysis.

Despite their recent advancements, Large Language Models (LLMs) still struggle to directly generate correct plans for complex multi-constraint planning problems, even with self-verification and self-critique. For example, a U.S. domestic travel planning benchmark TravelPlanner was proposed in Xie et al. (2024), where the best LLM OpenAI o1-preview can only find travel plans that satisfy user requirements with a 10% success rate given all needed information. In this work, we tackle this difficult problem by proposing an LLM-based planning framework that formalizes and solves complex multi-constraint planning problems as constrained satisfiability problems, which are further consumed by sound and complete satisfiability solvers. We start with TravelPlanner as the primary use case and achieve a success rate of 93.9%. We demonstrate our framework's robustness by showing its effectiveness in diverse paraphrased prompts. More importantly, our framework has strong zero-shot generalizability: It can successfully handle unseen constraints in a completely unseen international travel dataset we created, and it can even generalize well to new domains such as routing and task allocation problems in a zero-shot manner. Moreover, when user input queries are infeasible, our framework can identify the unsatisfiable core, provide failure reasons, and offers personalized modification suggestions to users according to diverse human preferences. We show that our framework can modify and solve for an average of 81.6% and 91.7% unsatisfiable queries from two datasets and prove with ablations that all key components of our framework are effective and necessary., Comment: 48 pages, 6 figures, 8 tables

Connected multi-agent robotic systems (MRS) are prone to deadlocks in an obstacle environment where the robots can get stuck away from their desired locations under a smooth low-level control policy. Without an external intervention, often in terms of a high-level command, a low-level control policy cannot resolve such deadlocks. Utilizing the generalizability and low data requirements of foundation models, this paper explores the possibility of using text-based models, i.e., large language models (LLMs), and text-and-image-based models, i.e., vision-language models (VLMs), as high-level planners for deadlock resolution. We propose a hierarchical control framework where a foundation model-based high-level planner helps to resolve deadlocks by assigning a leader to the MRS along with a set of waypoints for the MRS leader. Then, a low-level distributed control policy based on graph neural networks is executed to safely follow these waypoints, thereby evading the deadlock. We conduct extensive experiments on various MRS environments using the best available pre-trained LLMs and VLMs. We compare their performance with a graph-based planner in terms of effectiveness in helping the MRS reach their target locations and computational time. Our results illustrate that, compared to grid-based planners, the foundation models perform better in terms of the goal-reaching rate and computational time for complex environments, which helps us conclude that foundation models can assist MRS operating in complex obstacle-cluttered environments to resolve deadlocks efficiently.

Before autonomous systems can be deployed in safety-critical applications, we must be able to understand and verify the safety of these systems. For cases where the risk or cost of real-world testing is prohibitive, we propose a simulation-based framework for a) predicting ways in which an autonomous system is likely to fail and b) automatically adjusting the system's design and control policy to preemptively mitigate those failures. Existing tools for failure prediction struggle to search over high-dimensional environmental parameters, cannot efficiently handle end-to-end testing for systems with vision in the loop, and provide little guidance on how to mitigate failures once they are discovered. We approach this problem through the lens of approximate Bayesian inference and use differentiable simulation and rendering for efficient failure case prediction and repair. For cases where a differentiable simulator is not available, we provide a gradient-free version of our algorithm, and we include a theoretical and empirical evaluation of the trade-offs between gradient-based and gradient-free methods. We apply our approach on a range of robotics and control problems, including optimizing search patterns for robot swarms, UAV formation control, and robust network control. Compared to optimization-based falsification methods, our method predicts a more diverse, representative set of failure modes, and we find that our use of differentiable simulation yields solutions that have up to 10x lower cost and requires up to 2x fewer iterations to converge relative to gradient-free techniques. In hardware experiments, we find that repairing control policies using our method leads to a 5x robustness improvement. Accompanying code and video can be found at https://mit-realm.github.io/radium/

Sampling-based motion planning methods for manipulators in crowded environments often suffer from expensive collision checking and high sampling complexity, which make them difficult to use in real time. To address this issue, we propose a new generalizable control barrier function (CBF)-based steering controller to reduce the number of samples needed in a sampling-based motion planner RRT. Our method combines the strength of CBF for real-time collision-avoidance control and RRT for long-horizon motion planning, by using CBF-induced neural controller (CBF-INC) to generate control signals that steer the system towards sampled configurations by RRT. CBF-INC is learned as Neural Networks and has two variants handling different inputs, respectively: state (signed distance) input and point-cloud input from LiDAR. In the latter case, we also study two different settings: fully and partially observed environmental information. Compared to manually crafted CBF which suffers from over-approximating robot geometry, CBF-INC can balance safety and goal-reaching better without being over-conservative. Given state-based input, our neural CBF-induced neural controller-enhanced RRT (CBF-INC-RRT) can increase the success rate by 14% while reducing the number of nodes explored by 30%, compared with vanilla RRT on hard test cases. Given LiDAR input where vanilla RRT is not directly applicable, we demonstrate that our CBF-INC-RRT can improve the success rate by 10%, compared with planning with other steering controllers. Our project page with supplementary material is at https://mit-realm.github.io/CBF-INC-RRT-website/., Comment: Accepted by IEEE International Conference on Robotics and Automation (ICRA2024)

The tokamak offers a promising path to fusion energy, but plasma disruptions pose a major economic risk, motivating considerable advances in disruption avoidance. This work develops a reinforcement learning approach to this problem by training a policy to safely ramp-down the plasma current while avoiding limits on a number of quantities correlated with disruptions. The policy training environment is a hybrid physics and machine learning model trained on simulations of the SPARC primary reference discharge (PRD) ramp-down, an upcoming burning plasma scenario which we use as a testbed. To address physics uncertainty and model inaccuracies, the simulation environment is massively parallelized on GPU with randomized physics parameters during policy training. The trained policy is then successfully transferred to a higher fidelity simulator where it successfully ramps down the plasma while avoiding user-specified disruptive limits. We also address the crucial issue of safety criticality by demonstrating that a constraint-conditioned policy can be used as a trajectory design assistant to design a library of feed-forward trajectories to handle different physics conditions and user settings. As a library of trajectories is more interpretable and verifiable offline, we argue such an approach is a promising path for leveraging the capabilities of reinforcement learning in the safety-critical context of burning plasma tokamaks. Finally, we demonstrate how the training environment can be a useful platform for other feed-forward optimization approaches by using an evolutionary algorithm to perform optimization of feed-forward trajectories that are robust to physics uncertainty

Prompt optimization aims to find the best prompt to a large language model (LLM) for a given task. LLMs have been successfully used to help find and improve prompt candidates for single-step tasks. However, realistic tasks for agents are multi-step and introduce new challenges: (1) Prompt content is likely to be more extensive and complex, making it more difficult for LLMs to analyze errors, (2) the impact of an individual step is difficult to evaluate, and (3) different people may have varied preferences about task execution. While humans struggle to optimize prompts, they are good at providing feedback about LLM outputs; we therefore introduce a new LLM-driven discrete prompt optimization framework PRompt Optimization in Multi-Step Tasks (PROMST) that incorporates human-designed feedback rules to automatically offer direct suggestions for improvement. We also use an extra learned heuristic model that predicts prompt performance to efficiently sample from prompt candidates. This approach significantly outperforms both human-engineered prompts and several other prompt optimization methods across 11 representative multi-step tasks (an average 10.6\%-29.3\% improvement to current best methods on five LLMs respectively). We believe our work can serve as a benchmark for automatic prompt optimization for LLM-driven multi-step tasks. Datasets and Codes are available at https://github.com/yongchao98/PROMST. Project Page is available at https://yongchao98.github.io/MIT-REALM-PROMST., Comment: 62 pages, 14 figures, Published in EMNLP 2024 Main

Distributed, scalable, and safe control of large-scale multi-agent systems (MAS) is a challenging problem. In this paper, we design a distributed framework for safe multi-agent control in large-scale environments with obstacles, where a large number of agents are required to maintain safety using only local information and reach their goal locations. We introduce a new class of certificates, termed graph control barrier function (GCBF), which are based on the well-established control barrier function (CBF) theory for safety guarantees and utilize a graph structure for scalable and generalizable distributed control of MAS. We develop a novel theoretical framework to prove the safety of an arbitrary-sized MAS with a single GCBF. We propose a new training framework GCBF+ that uses graph neural networks (GNNs) to parameterize a candidate GCBF and a distributed control policy. The proposed framework is distributed and is capable of directly taking point clouds from LiDAR, instead of actual state information, for real-world robotic applications. We illustrate the efficacy of the proposed method through various hardware experiments on a swarm of drones with objectives ranging from exchanging positions to docking on a moving target without collision. Additionally, we perform extensive numerical experiments, where the number and density of agents, as well as the number of obstacles, increase. Empirical results show that in complex environments with nonlinear agents (e.g., Crazyflie drones) GCBF+ outperforms the handcrafted CBF-based method with the best performance by up to 20% for relatively small-scale MAS for up to 256 agents, and leading reinforcement learning (RL) methods by up to 40% for MAS with 1024 agents. Furthermore, the proposed method does not compromise on the performance, in terms of goal reaching, for achieving high safety rates, which is a common trade-off in RL-based methods., Comment: 18 pages, 12 figures, submitted to IEEE T-RO. arXiv admin note: text overlap with arXiv:2311.13014

The 2021 paper "Control barrier functions for stochastic systems" provides theorems that give almost sure safety guarantees given stochastic zero control barrier function (ZCBF). Unfortunately, both the theorem and its proof is invalid. In this letter, we illustrate on a toy example that the almost sure safety guarantees for stochastic ZCBF do not hold and explain why the proof is flawed. Although stochastic reciprocal barrier functions (RCBF) also uses the same proof technique, we provide a different proof technique that verifies that stochastic RCBFs are indeed safe with probability one. Using the RCBF, we derive a modified ZCBF condition that guarantees safety with probability one. Finally, we provide some discussion on the role of unbounded controls in the almost-sure safety guarantees of RCBFs, and show that the rate of divergence of the ratio of the drift and diffusion is the key for whether a system has almost sure safety guarantees., Comment: Under Review

In this survey, we review the recent advances in control design methods for robotic multi-agent systems (MAS), focussing on learning-based methods with safety considerations. We start by reviewing various notions of safety and liveness properties, and modeling frameworks used for problem formulation of MAS. Then we provide a comprehensive review of learning-based methods for safe control design for multi-robot systems. We start with various types of shielding-based methods, such as safety certificates, predictive filters, and reachability tools. Then, we review the current state of control barrier certificate learning in both a centralized and distributed manner, followed by a comprehensive review of multi-agent reinforcement learning with a particular focus on safety. Next, we discuss the state-of-the-art verification tools for the correctness of learning-based methods. Based on the capabilities and the limitations of the state of the art methods in learning and verification for MAS, we identify various broad themes for open challenges: how to design methods that can achieve good performance along with safety guarantees; how to decompose single-agent based centralized methods for MAS; how to account for communication-related practical issues; and how to assess transfer of theoretical guarantees to practice., Comment: Submitted to Annual Reviews in Control

We consider the problem of designing distributed collision-avoidance multi-agent control in large-scale environments with potentially moving obstacles, where a large number of agents are required to maintain safety using only local information and reach their goals. This paper addresses the problem of collision avoidance, scalability, and generalizability by introducing graph control barrier functions (GCBFs) for distributed control. The newly introduced GCBF is based on the well-established CBF theory for safety guarantees but utilizes a graph structure for scalable and generalizable decentralized control. We use graph neural networks to learn both neural a GCBF certificate and distributed control. We also extend the framework from handling state-based models to directly taking point clouds from LiDAR for more practical robotics settings. We demonstrated the efficacy of GCBF in a variety of numerical experiments, where the number, density, and traveling distance of agents, as well as the number of unseen and uncontrolled obstacles increase. Empirical results show that GCBF outperforms leading methods such as MAPPO and multi-agent distributed CBF (MDCBF). Trained with only 16 agents, GCBF can achieve up to 3 times improvement of success rate (agents reach goals and never encountered in any collisions) on <500 agents, and still maintain more than 50% success rates for >1000 agents when other methods completely fail., Comment: 20 pages, 10 figures; Accepted by 7th Conference on Robot Learning (CoRL 2023)

Control barrier functions (CBF) have become popular as a safety filter to guarantee the safety of nonlinear dynamical systems for arbitrary inputs. However, it is difficult to construct functions that satisfy the CBF constraints for high relative degree systems with input constraints. To address these challenges, recent work has explored learning CBFs using neural networks via neural CBF (NCBF). However, such methods face difficulties when scaling to higher dimensional systems under input constraints. In this work, we first identify challenges that NCBFs face during training. Next, to address these challenges, we propose policy neural CBF (PNCBF), a method of constructing CBFs by learning the value function of a nominal policy, and show that the value function of the maximum-over-time cost is a CBF. We demonstrate the effectiveness of our method in simulation on a variety of systems ranging from toy linear systems to an F-16 jet with a 16-dimensional state space. Finally, we validate our approach on a two-agent quadcopter system on hardware under tight input constraints., Comment: Submitted to ICRA 2024. Project page can be found at https://mit-realm.github.io/pncbf

To ensure safe, reliable operation of the electrical grid, we must be able to predict and mitigate likely failures. This need motivates the classic security-constrained AC optimal power flow (SCOPF) problem. SCOPF is commonly solved using adversarial optimization, where the dispatcher and an adversary take turns optimizing a robust dispatch and adversarial attack, respectively. We show that adversarial optimization is liable to severely overestimate the robustness of the optimized dispatch (when the adversary encounters a local minimum), leading the operator to falsely believe that their dispatch is secure. To prevent this overconfidence, we develop a novel adversarial sampling approach that prioritizes diversity in the predicted attacks. We find that our method not only substantially improves the robustness of the optimized dispatch but also avoids overconfidence, accurately characterizing the likelihood of voltage collapse under a given threat model. We demonstrate a proof-of-concept on small-scale transmission systems with 14 and 57 nodes., Comment: Accepted at NAPS 2023

A flurry of recent work has demonstrated that pre-trained large language models (LLMs) can be effective task planners for a variety of single-robot tasks. The planning performance of LLMs is significantly improved via prompting techniques, such as in-context learning or re-prompting with state feedback, placing new importance on the token budget for the context window. An under-explored but natural next direction is to investigate LLMs as multi-robot task planners. However, long-horizon, heterogeneous multi-robot planning introduces new challenges of coordination while also pushing up against the limits of context window length. It is therefore critical to find token-efficient LLM planning frameworks that are also able to reason about the complexities of multi-robot coordination. In this work, we compare the task success rate and token efficiency of four multi-agent communication frameworks (centralized, decentralized, and two hybrid) as applied to four coordination-dependent multi-agent 2D task scenarios for increasing numbers of agents. We find that a hybrid framework achieves better task success rates across all four tasks and scales better to more agents. We further demonstrate the hybrid frameworks in 3D simulations where the vision-to-text problem and dynamical errors are considered. See our project website https://yongchao98.github.io/MIT-REALM-Multi-Robot/ for prompts, videos, and code., Comment: 7 pages, 8 figures

Autonomous robotic systems, such as quadrotors, are susceptible to actuator faults, and for the safe operation of such systems, timely detection and isolation of these faults is essential. Neural networks can be used for verification of actuator performance via online actuator fault detection with high accuracy. In this paper, we develop a novel model-free fault detection and isolation (FDI) framework for quadrotor systems using long-short-term memory (LSTM) neural network architecture. The proposed framework only uses system output data and the commanded control input and requires no knowledge of the system model. Utilizing the symmetry in quadrotor dynamics, we train the FDI for fault in just one of the motors (e.g., motor $\# 2$), and the trained FDI can predict faults in any of the motors. This reduction in search space enables us to design an FDI for partial fault as well as complete fault scenarios. Numerical experiments illustrate that the proposed NN-FDI correctly verifies the actuator performance and identifies partial as well as complete faults with over $90\%$ prediction accuracy. We also illustrate that model-free NN-FDI performs at par with model-based FDI, and is robust to model uncertainties as well as distribution shifts in input data., Comment: Accepted for 2023 Allerton Conference on Communication, Control, & Computing

Before autonomous systems can be deployed in safety-critical applications, we must be able to understand and verify the safety of these systems. For cases where the risk or cost of real-world testing is prohibitive, we propose a simulation-based framework for a) predicting ways in which an autonomous system is likely to fail and b) automatically adjusting the system's design to preemptively mitigate those failures. We frame this problem through the lens of approximate Bayesian inference and use differentiable simulation for efficient failure case prediction and repair. We apply our approach on a range of robotics and control problems, including optimizing search patterns for robot swarms and reducing the severity of outages in power transmission networks. Compared to optimization-based falsification techniques, our method predicts a more diverse, representative set of failure modes, and we also find that our use of differentiable simulation yields solutions that have up to 10x lower cost and requires up to 2x fewer iterations to converge relative to gradient-free techniques. Code and videos can be found at https://mit-realm.github.io/breaking-things/, Comment: To appear at the 2023 Conference on Robot Learning (CoRL)

Ensuring safety and meeting temporal specifications are critical challenges for long-term robotic tasks. Signal temporal logic (STL) has been widely used to systematically and rigorously specify these requirements. However, traditional methods of finding the control policy under those STL requirements are computationally complex and not scalable to high-dimensional or systems with complex nonlinear dynamics. Reinforcement learning (RL) methods can learn the policy to satisfy the STL specifications via hand-crafted or STL-inspired rewards, but might encounter unexpected behaviors due to ambiguity and sparsity in the reward. In this paper, we propose a method to directly learn a neural network controller to satisfy the requirements specified in STL. Our controller learns to roll out trajectories to maximize the STL robustness score in training. In testing, similar to Model Predictive Control (MPC), the learned controller predicts a trajectory within a planning horizon to ensure the satisfaction of the STL requirement in deployment. A backup policy is designed to ensure safety when our controller fails. Our approach can adapt to various initial conditions and environmental parameters. We conduct experiments on six tasks, where our method with the backup policy outperforms the classical methods (MPC, STL-solver), model-free and model-based RL methods in STL satisfaction rate, especially on tasks with complex STL specifications while being 10X-100X faster than the classical methods., Comment: Accepted by IEEE Robotics and Automation Letters (RA-L) and ICRA2024

Designing stabilizing controllers is a fundamental challenge in autonomous systems, particularly for high-dimensional, nonlinear systems that can hardly be accurately modeled with differential equations. The Lyapunov theory offers a solution for stabilizing control systems, still, current methods relying on Lyapunov functions require access to complete dynamics or samples of system executions throughout the entire state space. Consequently, they are impractical for high-dimensional systems. This paper introduces a novel framework, LYapunov-Guided Exploration (LYGE), for learning stabilizing controllers tailored to high-dimensional, unknown systems. LYGE employs Lyapunov theory to iteratively guide the search for samples during exploration while simultaneously learning the local system dynamics, control policy, and Lyapunov functions. We demonstrate its scalability on highly complex systems, including a high-fidelity F-16 jet model featuring a 16D state space and a 4D input space. Experiments indicate that, compared to prior works in reinforcement learning, imitation learning, and neural certificates, LYGE reduces the distance to the goal by 50% while requiring only 5% to 32% of the samples. Furthermore, we demonstrate that our algorithm can be extended to learn controllers guided by other certificate functions for unknown systems., Comment: 32 pages, 7 figures; Accepted by the 6th Annual Conference on Learning for Dynamics and Control (L4DC 2024)

For effective human-robot interaction, robots need to understand, plan, and execute complex, long-horizon tasks described by natural language. Recent advances in large language models (LLMs) have shown promise for translating natural language into robot action sequences for complex tasks. However, existing approaches either translate the natural language directly into robot trajectories or factor the inference process by decomposing language into task sub-goals and relying on a motion planner to execute each sub-goal. When complex environmental and temporal constraints are involved, inference over planning tasks must be performed jointly with motion plans using traditional task-and-motion planning (TAMP) algorithms, making factorization into subgoals untenable. Rather than using LLMs to directly plan task sub-goals, we instead perform few-shot translation from natural language task descriptions to an intermediate task representation that can then be consumed by a TAMP algorithm to jointly solve the task and motion plan. To improve translation, we automatically detect and correct both syntactic and semantic errors via autoregressive re-prompting, resulting in significant improvements in task completion. We show that our approach outperforms several methods using LLMs as planners in complex task domains. See our project website https://yongchao98.github.io/MIT-REALM-AutoTAMP/ for prompts, videos, and code., Comment: 8 pages, 4 figures

Tasks for autonomous robotic systems commonly require stabilization to a desired region while maintaining safety specifications. However, solving this multi-objective problem is challenging when the dynamics are nonlinear and high-dimensional, as traditional methods do not scale well and are often limited to specific problem structures. To address this issue, we propose a novel approach to solve the stabilize-avoid problem via the solution of an infinite-horizon constrained optimal control problem (OCP). We transform the constrained OCP into epigraph form and obtain a two-stage optimization problem that optimizes over the policy in the inner problem and over an auxiliary variable in the outer problem. We then propose a new method for this formulation that combines an on-policy deep reinforcement learning algorithm with neural network regression. Our method yields better stability during training, avoids instabilities caused by saddle-point finding, and is not restricted to specific requirements on the problem structure compared to more traditional methods. We validate our approach on different benchmark tasks, ranging from low-dimensional toy examples to an F16 fighter jet with a 17-dimensional state space. Simulation results show that our approach consistently yields controllers that match or exceed the safety of existing methods while providing ten-fold increases in stability performance from larger regions of attraction., Comment: Accepted to Robotics: Science and Systems 2023. Project page can be found at https://mit-realm.github.io/efppo

Temporal Logic (TL) can be used to rigorously specify complex high-level specification for systems in many engineering applications. The translation between natural language (NL) and TL has been under-explored due to the lack of dataset and generalizable model across different application domains. In this paper, we propose an accurate and generalizable transformation framework of English instructions from NL to TL, exploring the use of Large Language Models (LLMs) at multiple stages. Our contributions are twofold. First, we develop a framework to create a dataset of NL-TL pairs combining LLMs and human annotation. We publish a dataset with 28K NL-TL pairs. Then, we finetune T5 models on the lifted versions (i.e., the specific Atomic Propositions (AP) are hidden) of the NL and TL. The enhanced generalizability originates from two aspects: 1) Usage of lifted NL-TL characterizes common logical structures, without constraints of specific domains. 2) Application of LLMs in dataset creation largely enhances corpus richness. We test the generalization of trained models on five varied domains. To achieve full NL-TL transformation, we either combine the lifted model with AP recognition task or do the further finetuning on each specific domain. During the further finetuning, our model achieves higher accuracy (>95%) using only <10% training data, compared with the baseline sequence to sequence (Seq2Seq) model., Comment: 25 pages, 18 figures

Developing stable controllers for large-scale networked dynamical systems is crucial but has long been challenging due to two key obstacles: certifiability and scalability. In this paper, we present a general framework to solve these challenges using compositional neural certificates based on ISS (Input-to-State Stability) Lyapunov functions. Specifically, we treat a large networked dynamical system as an interconnection of smaller subsystems and develop methods that can find each subsystem a decentralized controller and an ISS Lyapunov function; the latter can be collectively composed to prove the global stability of the system. To ensure the scalability of our approach, we develop generalizable and robust ISS Lyapunov functions where a single function can be used across different subsystems and the certificates we produced for small systems can be generalized to be used on large systems with similar structures. We encode both ISS Lyapunov functions and controllers as neural networks and propose a novel training methodology to handle the logic in ISS Lyapunov conditions that encodes the interconnection with neighboring subsystems. We demonstrate our approach in systems including Platoon, Drone formation control, and Power systems. Experimental results show that our framework can reduce the tracking error up to 75% compared with RL algorithms when applied to large-scale networked systems., Comment: 25 pages, 8 figures; Accepted by 5th Annual Learning for Dynamics & Control Conference (L4DC) 2023

Hybrid systems are prevalent in robotics. However, ensuring the stability of hybrid systems is challenging due to sophisticated continuous and discrete dynamics. A system with all its system modes stable can still be unstable. Hence special treatments are required at mode switchings to stabilize the system. In this work, we propose a hierarchical, neural network (NN)-based method to control general hybrid systems. For each system mode, we first learn an NN Lyapunov function and an NN controller to ensure the states within the region of attraction (RoA) can be stabilized. Then an RoA NN estimator is learned across different modes. Upon mode switching, we propose a differentiable planner to ensure the states after switching can land in next mode's RoA, hence stabilizing the hybrid system. We provide novel theoretical stability guarantees and conduct experiments in car tracking control, pogobot navigation, and bipedal walker locomotion. Our method only requires 0.25X of the training time as needed by other learning-based methods. With low running time (10-50X faster than model predictive control (MPC)), our controller achieves a higher stability/success rate over other baselines such as MPC, reinforcement learning (RL), common Lyapunov methods (CLF), linear quadratic regulator (LQR), quadratic programming (QP) and Hamilton-Jacobian-based methods (HJB). The project page is on https://mit-realm.github.io/hybrid-clf., Comment: Accepted to L4DC2023

Large-scale self-supervised models have recently revolutionized our ability to perform a variety of tasks within the vision and language domains. However, using such models for autonomous systems is challenging because of safety requirements: besides executing correct actions, an autonomous agent must also avoid the high cost and potentially fatal critical mistakes. Traditionally, self-supervised training mainly focuses on imitating previously observed behaviors, and the training demonstrations carry no notion of which behaviors should be explicitly avoided. In this work, we propose Control Barrier Transformer (ConBaT), an approach that learns safe behaviors from demonstrations in a self-supervised fashion. ConBaT is inspired by the concept of control barrier functions in control theory and uses a causal transformer that learns to predict safe robot actions autoregressively using a critic that requires minimal safety data labeling. During deployment, we employ a lightweight online optimization to find actions that ensure future states lie within the learned safe set. We apply our approach to different simulated control tasks and show that our method results in safer control policies compared to other classical and learning-based methods such as imitation learning, reinforcement learning, and model predictive control.

Model Predictive Path Integral (MPPI) control is a type of sampling-based model predictive control that simulates thousands of trajectories and uses these trajectories to synthesize optimal controls on-the-fly. In practice, however, MPPI encounters problems limiting its application. For instance, it has been observed that MPPI tends to make poor decisions if unmodeled dynamics or environmental disturbances exist, preventing its use in safety-critical applications. Moreover, the multi-threaded simulations used by MPPI require significant onboard computational resources, making the algorithm inaccessible to robots without modern GPUs. To alleviate these issues, we propose a novel (Shield-MPPI) algorithm that provides robustness against unpredicted disturbances and achieves real-time planning using a much smaller number of parallel simulations on regular CPUs. The novel Shield-MPPI algorithm is tested on an aggressive autonomous racing platform both in simulation and using experiments. The results show that the proposed controller greatly reduces the number of constraint violations compared to state-of-the-art robust MPPI variants and stochastic MPC methods., Comment: 8 pages, 7 figures. Submitted to RA-L for review

Learning-based methods have shown promising performance for accelerating motion planning, but mostly in the setting of static environments. For the more challenging problem of planning in dynamic environments, such as multi-arm assembly tasks and human-robot interaction, motion planners need to consider the trajectories of the dynamic obstacles and reason about temporal-spatial interactions in very large state spaces. We propose a GNN-based approach that uses temporal encoding and imitation learning with data aggregation for learning both the embeddings and the edge prioritization policies. Experiments show that the proposed methods can significantly accelerate online planning over state-of-the-art complete dynamic planning algorithms. The learned models can often reduce costly collision checking operations by more than 1000x, and thus accelerating planning by up to 95%, while achieving high success rates on hard instances as well.

Uncertainty is prevalent in robotics. Due to measurement noise and complex dynamics, we cannot estimate the exact system and environment state. Since conservative motion planners are not guaranteed to find a safe control strategy in a crowded, uncertain environment, we propose a density-based method. Our approach uses a neural network and the Liouville equation to learn the density evolution for a system with an uncertain initial state. We can plan for feasible and probably safe trajectories by applying a gradient-based optimization procedure to minimize the collision risk. We conduct motion planning experiments on simulated environments and environments generated from real-world data and outperform baseline methods such as model predictive control and nonlinear programming. While our method requires offline planning, the online run time is 100 times smaller compared to model predictive control., Comment: \copyright 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

In order to be effective partners for humans, robots must become increasingly comfortable with making contact with their environment. Unfortunately, it is hard for robots to distinguish between ``just enough'' and ``too much'' force: some force is required to accomplish the task but too much might damage equipment or injure humans. Traditional approaches to designing compliant force-feedback controllers, such as stiffness control, require difficult hand-tuning of control parameters and make it difficult to build safe, effective robot collaborators. In this paper, we propose a novel yet easy-to-implement force feedback controller that uses control barrier functions (CBFs) to derive a compliant controller directly from users' specifications of the maximum allowable forces and torques. We compare our approach to traditional stiffness control to demonstrate potential advantages of our control architecture, and we demonstrate the effectiveness of our controller on an example human-robot collaboration task: cooperative manipulation of a bulky object.

To navigate complex environments, robots must increasingly use high-dimensional visual feedback (e.g. images) for control. However, relying on high-dimensional image data to make control decisions raises important questions; particularly, how might we prove the safety of a visual-feedback controller? Control barrier functions (CBFs) are powerful tools for certifying the safety of feedback controllers in the state-feedback setting, but CBFs have traditionally been poorly-suited to visual feedback control due to the need to predict future observations in order to evaluate the barrier function. In this work, we solve this issue by leveraging recent advances in neural radiance fields (NeRFs), which learn implicit representations of 3D scenes and can render images from previously-unseen camera perspectives, to provide single-step visual foresight for a CBF-based controller. This novel combination is able to filter out unsafe actions and intervene to preserve safety. We demonstrate the effect of our controller in real-time simulation experiments where it successfully prevents the robot from taking dangerous actions., Comment: Accepted to ICRA 2023

Density of the reachable states can help understand the risk of safety-critical systems, especially in situations when worst-case reachability is too conservative. Recent work provides a data-driven approach to compute the density distribution of autonomous systems' forward reachable states online. In this paper, we study the use of such approach in combination with model predictive control for verifiable safe path planning under uncertainties. We first use the learned density distribution to compute the risk of collision online. If such risk exceeds the acceptable threshold, our method will plan for a new path around the previous trajectory, with the risk of collision below the threshold. Our method is well-suited to handle systems with uncertainties and complicated dynamics as our data-driven approach does not need an analytical form of the systems' dynamics and can estimate forward state density with an arbitrary initial distribution of uncertainties. We design two challenging scenarios (autonomous driving and hovercraft control) for safe motion planning in environments with obstacles under system uncertainties. We first show that our density estimation approach can reach a similar accuracy as the Monte-Carlo-based method while using only 0.01X training samples. By leveraging the estimated risk, our algorithm achieves the highest success rate in goal reaching when enforcing the safety rate above 0.99., Comment: NASA Formal Methods 2022

There is a growing need for computational tools to automatically design and verify autonomous systems, especially complex robotic systems involving perception, planning, control, and hardware in the autonomy stack. Differentiable programming has recently emerged as powerful tool for modeling and optimization. However, very few studies have been done to understand how differentiable programming can be used for robust, certifiable end-to-end design optimization. In this paper, we fill this gap by combining differentiable programming for robot design optimization with a novel statistical framework for certifying the robustness of optimized designs. Our framework can conduct end-to-end optimization and robustness certification for robotics systems, enabling simultaneous optimization of navigation, perception, planning, control, and hardware subsystems. Using simulation and hardware experiments, we show how our tool can be used to solve practical problems in robotics. First, we optimize sensor placements for robot navigation (a design with 5 subsystems and 6 tunable parameters) in under 5 minutes to achieve an 8.4x performance improvement compared to the initial design. Second, we solve a multi-agent collaborative manipulation task (3 subsystems and 454 parameters) in under an hour to achieve a 44% performance improvement over the initial design. We find that differentiable programming enables much faster (32% and 20x, respectively for each example) optimization than approximate gradient methods. We certify the robustness of each design and successfully deploy the optimized designs in hardware. An open-source implementation is available at https://github.com/MIT-REALM/architect, Comment: Accepted for RSS 2022

Multi-robot assembly systems are becoming increasingly appealing in manufacturing due to their ability to automatically, flexibly, and quickly construct desired structural designs. However, effectively planning for these systems in a manner that ensures each robot is simultaneously productive, and not idle, is challenging due to (1) the close proximity that the robots must operate in to manipulate the structure and (2) the inherent structural partial orderings on when each part can be installed. In this paper, we present a task and motion planning framework that jointly plans safe, low-makespan plans for a team of robots to assemble complex spatial structures. Our framework takes a hierarchical approach that, at the high level, uses Mixed-integer Linear Programs to compute an abstract plan comprised of an allocation of robots to tasks subject to precedence constraints and, at the low level, builds on a state-of-the-art algorithm for Multi-Agent Path Finding to plan collision-free robot motions that realize this abstract plan. Critical to our approach is the inclusion of certain collision constraints and movement durations during high-level planning, which better informs the search for abstract plans that are likely to be both feasible and low-makespan while keeping the search tractable. We demonstrate our planning system on several challenging assembly domains with several (sometimes heterogeneous) robots with grippers or suction plates for assembling structures with up to 23 objects involving Lego bricks, bars, plates, or irregularly shaped blocks., Comment: 8 pages, 6 figures, 1 table

Signal temporal logic (STL) provides a powerful, flexible framework for specifying complex autonomy tasks; however, existing methods for planning based on STL specifications have difficulty scaling to long-horizon tasks and are not robust to external disturbances. In this paper, we present an algorithm for finding robust plans that satisfy STL specifications. Our method alternates between local optimization and local falsification, using automatically differentiable temporal logic to iteratively optimize its plan in response to counterexamples found during the falsification process. We benchmark our counterexample-guided planning method against state-of-the-art planning methods on two long-horizon satellite rendezvous missions, showing that our method finds high-quality plans that satisfy STL specifications despite adversarial disturbances. We find that our method consistently finds plans that are robust to adversarial disturbances and requires less than half the time of competing methods. We provide an implementation of our planner at https://github.com/MIT-REALM/architect.

Learning-enabled control systems have demonstrated impressive empirical performance on challenging control problems in robotics, but this performance comes at the cost of reduced transparency and lack of guarantees on the safety or stability of the learned controllers. In recent years, new techniques have emerged to provide these guarantees by learning certificates alongside control policies -- these certificates provide concise, data-driven proofs that guarantee the safety and stability of the learned control system. These methods not only allow the user to verify the safety of a learned controller but also provide supervision during training, allowing safety and stability requirements to influence the training process itself. In this paper, we provide a comprehensive survey of this rapidly developing field of certificate learning. We hope that this paper will serve as an accessible introduction to the theory and practice of certificate learning, both to those who wish to apply these tools to practical robotics problems and to those who wish to dive more deeply into the theory of learning for control., Comment: Accepted at IEEE Transactions on Robotics. Supplementary code available at https://github.com/MIT-REALM/neural_clbf

We tackle the challenging problem of multi-agent cooperative motion planning for complex tasks described using signal temporal logic (STL), where robots can have nonlinear and nonholonomic dynamics. Existing methods in multi-agent motion planning, especially those based on discrete abstractions and model predictive control (MPC), suffer from limited scalability with respect to the complexity of the task, the size of the workspace, and the planning horizon. We present a method based on {\em timed waypoints\/} to address this issue. We show that timed waypoints can help abstract nonlinear behaviors of the system as safety envelopes around the reference path defined by those waypoints. Then the search for waypoints satisfying the STL specifications can be inductively encoded as a mixed-integer linear program. The agents following the synthesized timed waypoints have their tasks automatically allocated, and are guaranteed to satisfy the STL specifications while avoiding collisions. We evaluate the algorithm on a wide variety of benchmarks. Results show that it supports multi-agent planning from complex specification over long planning horizons, and significantly outperforms state-of-the-art abstraction-based and MPC-based motion planning methods. The implementation is available at https://github.com/sundw2014/STLPlanning., Comment: Accepted to IEEE Robotics and Automation Letters (RA-L)

Control certificates based on barrier functions have been a powerful tool to generate probably safe control policies for dynamical systems. However, existing methods based on barrier certificates are normally for white-box systems with differentiable dynamics, which makes them inapplicable to many practical applications where the system is a black-box and cannot be accurately modeled. On the other side, model-free reinforcement learning (RL) methods for black-box systems suffer from lack of safety guarantees and low sampling efficiency. In this paper, we propose a novel method that can learn safe control policies and barrier certificates for black-box dynamical systems, without requiring for an accurate system model. Our method re-designs the loss function to back-propagate gradient to the control policy even when the black-box dynamical system is non-differentiable, and we show that the safety certificates hold on the black-box system. Empirical results in simulation show that our method can significantly improve the performance of the learned policies by achieving nearly 100% safety and goal reaching rates using much fewer training samples, compared to state-of-the-art black-box safe control methods. Our learned agents can also generalize to unseen scenarios while keeping the original performance. The source code can be found at https://github.com/Zengyi-Qin/bcbf., Comment: IEEE Robotics and Automation Letters, 2022

Many robotic tasks require high-dimensional sensors such as cameras and Lidar to navigate complex environments, but developing certifiably safe feedback controllers around these sensors remains a challenging open problem, particularly when learning is involved. Previous works have proved the safety of perception-feedback controllers by separating the perception and control subsystems and making strong assumptions on the abilities of the perception subsystem. In this work, we introduce a novel learning-enabled perception-feedback hybrid controller, where we use Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) to show the safety and liveness of a full-stack perception-feedback controller. We use neural networks to learn a CBF and CLF for the full-stack system directly in the observation space of the robot, without the need to assume a separate perception-based state estimator. Our hybrid controller, called LOCUS (Learning-enabled Observation-feedback Control Using Switching), can safely navigate unknown environments, consistently reach its goal, and generalizes safely to environments outside of the training dataset. We demonstrate LOCUS in experiments both in simulation and in hardware, where it successfully navigates a changing environment using feedback from a Lidar sensor., Comment: Accepted for publication in RA-L

This paper presents a theoretical overview of a Neural Contraction Metric (NCM): a neural network model of an optimal contraction metric and corresponding differential Lyapunov function, the existence of which is a necessary and sufficient condition for incremental exponential stability of non-autonomous nonlinear system trajectories. Its innovation lies in providing formal robustness guarantees for learning-based control frameworks, utilizing contraction theory as an analytical tool to study the nonlinear stability of learned systems via convex optimization. In particular, we rigorously show in this paper that, by regarding modeling errors of the learning schemes as external disturbances, the NCM control is capable of obtaining an explicit bound on the distance between a time-varying target trajectory and perturbed solution trajectories, which exponentially decreases with time even under the presence of deterministic and stochastic perturbation. These useful features permit simultaneous synthesis of a contraction metric and associated control law by a neural network, thereby enabling real-time computable and probably robust learning-based control for general control-affine nonlinear systems., Comment: IEEE Conference on Decision and Control (CDC), Preprint Version. Accepted July, 2021

State density distribution, in contrast to worst-case reachability, can be leveraged for safety-related problems to better quantify the likelihood of the risk for potentially hazardous situations. In this work, we propose a data-driven method to compute the density distribution of reachable states for nonlinear and even black-box systems. Our semi-supervised approach learns system dynamics and the state density jointly from trajectory data, guided by the fact that the state density evolution follows the Liouville partial differential equation. With the help of neural network reachability tools, our approach can estimate the set of all possible future states as well as their density. Moreover, we could perform online safety verification with probability ranges for unsafe behaviors to occur. We use an extensive set of experiments to show that our learned solution can produce a much more accurate estimate on density distribution, and can quantify risks less conservatively and flexibly comparing with worst-case analysis., Comment: Accepted at CoRL 2021

Safety and stability are common requirements for robotic control systems; however, designing safe, stable controllers remains difficult for nonlinear and uncertain models. We develop a model-based learning approach to synthesize robust feedback controllers with safety and stability guarantees. We take inspiration from robust convex optimization and Lyapunov theory to define robust control Lyapunov barrier functions that generalize despite model uncertainty. We demonstrate our approach in simulation on problems including car trajectory tracking, nonlinear control with obstacle avoidance, satellite rendezvous with safety constraints, and flight control with a learned ground effect model. Simulation results show that our approach yields controllers that match or exceed the capabilities of robust MPC while reducing computational costs by an order of magnitude., Comment: Accepted to the 5th Annual Conference on Robot Learning (CoRL 21)

Reactive and safe agent modelings are important for nowadays traffic simulator designs and safe planning applications. In this work, we proposed a reactive agent model which can ensure safety without comprising the original purposes, by learning only high-level decisions from expert data and a low-level decentralized controller guided by the jointly learned decentralized barrier certificates. Empirical results show that our learned road user simulation models can achieve a significant improvement in safety comparing to state-of-the-art imitation learning and pure control-based methods, while being similar to human agents by having smaller errors to the expert data. Moreover, our learned reactive agents are shown to generalize better to unseen traffic conditions, and react better to other road users and therefore can help understand challenging planning problems pragmatically., Comment: Accepted at IROS 2021

We study constrained reinforcement learning (CRL) from a novel perspective by setting constraints directly on state density functions, rather than the value functions considered by previous works. State density has a clear physical and mathematical interpretation, and is able to express a wide variety of constraints such as resource limits and safety requirements. Density constraints can also avoid the time-consuming process of designing and tuning cost functions required by value function-based constraints to encode system specifications. We leverage the duality between density functions and Q functions to develop an effective algorithm to solve the density constrained RL problem optimally and the constrains are guaranteed to be satisfied. We prove that the proposed algorithm converges to a near-optimal solution with a bounded error even when the policy update is imperfect. We use a set of comprehensive experiments to demonstrate the advantages of our approach over state-of-the-art CRL methods, with a wide range of density constrained tasks as well as standard CRL benchmarks such as Safety-Gym., Comment: Accepted by ICML, 2021

In this paper, we consider the problem of using a robot to explore an environment with an unknown, state-dependent disturbance function while avoiding some forbidden areas. The goal of the robot is to safely collect observations of the disturbance and construct an accurate estimate of the underlying disturbance function. We use Gaussian Process (GP) to get an estimate of the disturbance from data with a high-confidence bound on the regression error. Furthermore, we use neural Contraction Metrics to derive a tracking controller and the corresponding high-confidence uncertainty tube around the nominal trajectory planned for the robot, based on the estimate of the disturbance. From the robustness of the Contraction Metric, error bound can be pre-computed and used by the motion planner such that the actual trajectory is guaranteed to be safe. As the robot collects more and more observations along its trajectory, the estimate of the disturbance becomes more and more accurate, which in turn improves the performance of the tracking controller and enlarges the free space that the robot can safely explore. We evaluate the proposed method using a carefully designed environment with a ground vehicle. Results show that with the proposed method the robot can thoroughly explore the environment safely and quickly.