Search

AI advancements have been significantly driven by a combination of foundation models and curiosity-driven learning aimed at increasing capability and adaptability. A growing area of interest within this field is Open-Endedness - the ability of AI systems to continuously and autonomously generate novel and diverse artifacts or solutions. This has become relevant for accelerating scientific discovery and enabling continual adaptation in AI agents. This position paper argues that the inherently dynamic and self-propagating nature of Open-Ended AI introduces significant, underexplored risks, including challenges in maintaining alignment, predictability, and control. This paper systematically examines these challenges, proposes mitigation strategies, and calls for action for different stakeholders to support the safe, responsible and successful development of Open-Ended AI., Comment: 12 pages

Selective plane illumination microscopy (SPIM), also known as light sheet fluorescence microscopy, provides high specificity through fluorescence labeling. However, it lacks complementary structural information from the surrounding context, which is essential for the comprehensive analysis of biological samples. Here, we present a high-resolution, multimodal imaging system that integrates SPIM with full-field optical coherence tomography (FF-OCT), without requiring modifications to the existing SPIM setup. Both SPIM and FF-OCT offer low phototoxicity and intrinsic optical sectioning, making them well-suited for in vivo imaging. Their shared detection path enables seamless and efficient co-registration of fluorescence and structural data. We demonstrate the functionality of this combined system by performing in vivo imaging of zebrafish larvae., Comment: 9 pages, 5 figures

Climate change detection and attribution (D&A) is concerned with determining the extent to which anthropogenic activities have influenced specific aspects of the global climate system. D&A fits within the broader field of causal inference, the collection of statistical methods that identify cause and effect relationships. There are a wide variety of methods for making attribution statements, each of which require different types of input data and focus on different types of weather and climate events and each of which are conditional to varying extents. Some methods are based on Pearl causality while others leverage Granger causality, and the causal framing provides important context for how the resulting attribution conclusion should be interpreted. However, while Granger-causal attribution analyses have become more common, there is no clear statement of their strengths and weaknesses and no clear consensus on where and when Granger-causal perspectives are appropriate. In this prospective paper, we provide a formal definition for Granger-based approaches to trend and event attribution and a clear comparison with more traditional methods for assessing the human influence on extreme weather and climate events. Broadly speaking, Granger-causal attribution statements can be constructed quickly from observations and do not require computationally-intesive dynamical experiments. These analyses also enable rapid attribution, which is useful in the aftermath of a severe weather event, and provide multiple lines of evidence for anthropogenic climate change when paired with Pearl-causal attribution. Confidence in attribution statements is increased when different methodologies arrive at similar conclusions. Moving forward, we encourage the D&A community to embrace hybrid approaches to climate change attribution that leverage the strengths of both Granger and Pearl causality.

The last decade has seen numerous record-shattering heatwaves in all corners of the globe. In the aftermath of these devastating events, there is interest in identifying worst-case thresholds or upper bounds that quantify just how hot temperatures can become. Generalized Extreme Value theory provides a data-driven estimate of extreme thresholds; however, upper bounds may be exceeded by future events, which undermines attribution and planning for heatwave impacts. Here, we show how the occurrence and relative probability of observed events that exceed a priori upper bound estimates, so-called "impossible" temperatures, has changed over time. We find that many unprecedented events are actually within data-driven upper bounds, but only when using modern spatial statistical methods. Furthermore, there are clear connections between anthropogenic forcing and the "impossibility" of the most extreme temperatures. Robust understanding of heatwave thresholds provides critical information about future record-breaking events and how their extremity relates to historical measurements.

A plasma scenario with negative triangularity (NT) shaping is achieved on MAST-U for the first time. While edge localized modes (ELMs) are eventually suppressed as the triangularity is decreased below $\delta$ < -0.06, an extended period of H-mode operation with Type-III ELMs is sustained at less negative $\delta$ even through access to the second stability region for ideal ballooning modes is closed. This documents a qualitative difference from the ELM-free access conditions documented in NT scenarios on conventional aspect ratio machines. The electron temperature at the pedestal top drops across the transition to ELM-free operation, but a steady rise in core temperature as $\delta$ is decreased allows for similar normalized beta in the ELM-free NT and H-mode positive triangularity shapes.

The goal of future quantum networks is to enable new internet applications that are impossible to achieve using solely classical communication. Up to now, demonstrations of quantum network applications and functionalities on quantum processors have been performed in ad-hoc software that was specific to the experimental setup, programmed to perform one single task (the application experiment) directly into low-level control devices using expertise in experimental physics. Here, we report on the design and implementation of the first architecture capable of executing quantum network applications on quantum processors in platform-independent high-level software. We demonstrate the architecture's capability to execute applications in high-level software, by implementing it as a quantum network operating system -- QNodeOS -- and executing test programs including a delegated computation from a client to a server on two quantum network nodes based on nitrogen-vacancy (NV) centers in diamond. We show how our architecture allows us to maximize the use of quantum network hardware, by multitasking different applications on a quantum network for the first time. Our architecture can be used to execute programs on any quantum processor platform corresponding to our system model, which we illustrate by demonstrating an additional driver for QNodeOS for a trapped-ion quantum network node based on a single $^{40}\text{Ca}^+$ atom. Our architecture lays the groundwork for computer science research in the domain of quantum network programming, and paves the way for the development of software that can bring quantum network technology to society., Comment: 12 pages, 5 figures, supplementary materials (48 pages, 24 figures, 11 tables)

Previous simulations of cataclysmic variables studied either the quiescence, or the outburst state in multiple dimensions or they simulated complete outburst cycles in one dimension using simplified models for the gravitational torques. We self-consistently simulate complete outburst cycles of normal and superoutbursts in cataclysmic variable systems in two dimensions. We study the effect of different $\alpha$ viscosity parameters, mass transfer rates, and binary mass ratios on the disk luminosities, outburst occurrence rates, and superhumps. We simulate non-isothermal, viscous accretion disks in cataclysmic variable systems using a modified version of the FARGO code with an updated equation of state and a cooling function designed to reproduce s-curve behavior. Our simulations can model complete outburst cycles using the thermal tidal instability model. We find higher superhump amplitudes and stronger gravitational torques than previous studies, resulting in better agreement with observations., Comment: Accepted in A&A. 21 pages, 27 figures

Root Cause Analysis (RCA) is a quality management method that aims to systematically investigate and identify the cause-and-effect relationships of problems and their underlying causes. Traditional methods are based on the analysis of problems by subject matter experts. In modern production processes, large amounts of data are collected. For this reason, increasingly computer-aided and data-driven methods are used for RCA. One of these methods are Causal Discovery Algorithms (CDA). This publication demonstrates the application of CDA on data from the assembly of a leading automotive manufacturer. The algorithms used learn the causal structure between the characteristics of the manufactured vehicles, the ergonomics and the temporal scope of the involved assembly processes, and quality-relevant product features based on representative data. This publication compares various CDAs in terms of their suitability in the context of quality management. For this purpose, the causal structures learned by the algorithms as well as their runtime are compared. This publication provides a contribution to quality management and demonstrates how CDAs can be used for RCA in assembly processes., Comment: in German language

Network Utility Maximisation (NUM) addresses the problem of allocating resources fairly within a network and explores the ways to achieve optimal allocation in real-world networks. Although extensively studied in classical networks, NUM is an emerging area of research in the context of quantum networks. In this work, we consider the quantum network utility maximisation (QNUM) problem in a static setting, where a user's utility takes into account the assigned quantum quality (fidelity) via a generic entanglement measure as well as the corresponding rate of entanglement generation. Under certain assumptions, we demonstrate that the QNUM problem can be formulated as an optimisation problem with the rate allocation vector as the only decision variable. Using a change of variable technique known in the field of geometric programming, we then establish sufficient conditions under which this formulation can be reduced to a convex problem, a class of optimisation problems that can be solved efficiently and with certainty even in high dimensions. We further show that this technique preserves convexity, enabling us to formulate convex QNUM problems in networks where some routes have certain entanglement measures that do not readily admit convex formulation, while others do. This allows us to compute the optimal resource allocation in networks where heterogeneous applications run over different routes.

Learning to lift and rotate objects with the fingertips is necessary for autonomous in-hand dexterous manipulation. In our study, we explore the impact of various factors on successful learning strategies for this task. Specifically, we investigate the role of curriculum learning and haptic feedback in enabling the learning of dexterous manipulation. Using model-free Reinforcement Learning, we compare different curricula and two haptic information modalities (No-tactile vs. 3D-force sensing) for lifting and rotating a ball against gravity with a three-fingered simulated robotic hand with no visual input. Note that our best results were obtained when we used a novel curriculum-based learning rate scheduler, which adjusts the linearly-decaying learning rate when the reward is changed as it accelerates convergence to higher rewards. Our findings demonstrate that the choice of curriculum greatly biases the acquisition of different features of dexterous manipulation. Surprisingly, successful learning can be achieved even in the absence of tactile feedback, challenging conventional assumptions about the necessity of haptic information for dexterous manipulation tasks. We demonstrate the generalizability of our results to balls of different weights and sizes, underscoring the robustness of our learning approach. This work, therefore, emphasizes the importance of the choice curriculum and challenges long-held notions about the need for tactile information to autonomously learn in-hand dexterous manipulation.

Failure mode and effects analysis (FMEA) is a critical tool for mitigating potential failures, particular during ramp-up phases of new products. However, its effectiveness is often limited by the missing reasoning capabilities of the FMEA tools, which are usually tabular structured. Meanwhile, large language models (LLMs) offer novel prospects for fine-tuning on custom datasets for reasoning within FMEA contexts. However, LLMs face challenges in tasks that require factual knowledge, a gap that retrieval-augmented generation (RAG) approaches aim to fill. RAG retrieves information from a non-parametric data store and uses a language model to generate responses. Building on this idea, we propose to advance the non-parametric data store with a knowledge graph (KG). By enhancing the RAG framework with a KG, our objective is to leverage analytical and semantic question-answering capabilities on FMEA data. This paper contributes by presenting a new ontology for FMEA observations, an algorithm for creating vector embeddings from the FMEA KG, and a KG enhanced RAG framework. Our approach is validated through a human study and we measure the performance of the context retrieval recall and precision.

This paper introduces a post-hoc explainable AI method tailored for Knowledge Graph Embedding models. These models are essential to Knowledge Graph Completion yet criticized for their opaque, black-box nature. Despite their significant success in capturing the semantics of knowledge graphs through high-dimensional latent representations, their inherent complexity poses substantial challenges to explainability. Unlike existing methods, our approach directly decodes the latent representations encoded by Knowledge Graph Embedding models, leveraging the principle that similar embeddings reflect similar behaviors within the Knowledge Graph. By identifying distinct structures within the subgraph neighborhoods of similarly embedded entities, our method identifies the statistical regularities on which the models rely and translates these insights into human-understandable symbolic rules and facts. This bridges the gap between the abstract representations of Knowledge Graph Embedding models and their predictive outputs, offering clear, interpretable insights. Key contributions include a novel post-hoc explainable AI method for Knowledge Graph Embedding models that provides immediate, faithful explanations without retraining, facilitating real-time application even on large-scale knowledge graphs. The method's flexibility enables the generation of rule-based, instance-based, and analogy-based explanations, meeting diverse user needs. Extensive evaluations show our approach's effectiveness in delivering faithful and well-localized explanations, enhancing the transparency and trustworthiness of Knowledge Graph Embedding models.

Distributing quantum states reliably among distant locations is a key challenge in the field of quantum networks. One-way quantum networks address this by using one-way communication and quantum error correction. Here, we analyze quantum circuit switching as a protocol to distribute quantum states in one-way quantum networks. In quantum circuit switching, pairs of users can request the delivery of multiple quantum states from one user to the other. After waiting for approval from the network, the states can be distributed either sequentially, forwarding one at a time along a path of quantum repeaters, or in parallel, sending batches of quantum states from repeater to repeater. Since repeaters can only forward a finite number of quantum states at a time, a pivotal question arises: is it advantageous to send them sequentially (allowing for multiple requests simultaneously) or in parallel (reducing processing time but handling only one request at a time)? We compare both approaches in a quantum network with a star topology. Using tools from queuing theory, we show that requests are met at a higher rate when packets are distributed in parallel, although sequential distribution can generally provide service to a larger number of users simultaneously. We also show that using a large number of quantum repeaters to combat channel losses limits the maximum distance between users, as each repeater introduces additional processing delays. These findings provide insight into the design of protocols for distributing quantum states in one-way quantum networks., Comment: Main text: 9 pages, 5 figures. Appendices: 14 pages, 8 figures

Releasing open-source large language models (LLMs) presents a dual-use risk since bad actors can easily fine-tune these models for harmful purposes. Even without the open release of weights, weight stealing and fine-tuning APIs make closed models vulnerable to harmful fine-tuning attacks (HFAs). While safety measures like preventing jailbreaks and improving safety guardrails are important, such measures can easily be reversed through fine-tuning. In this work, we propose Representation Noising (RepNoise), a defence mechanism that operates even when attackers have access to the weights. RepNoise works by removing information about harmful representations such that it is difficult to recover them during fine-tuning. Importantly, our defence is also able to generalize across different subsets of harm that have not been seen during the defence process as long as they are drawn from the same distribution of the attack set. Our method does not degrade the general capability of LLMs and retains the ability to train the model on harmless tasks. We provide empirical evidence that the efficacy of our defence lies in its ``depth'': the degree to which information about harmful representations is removed across all layers of the LLM. We also find areas where RepNoise still remains ineffective and highlight how those limitations can inform future research., Comment: Published in NeurIPs 2024

The magnetic dipole and the spectroscopic quadrupole moments of the nuclear ground states in the odd-mass nickel isotopes $^{59-67}$Ni have been determined using collinear laser spectroscopy at the CERN-ISOLDE facility. They are compared to ab initio valence-space in-medium similarity renormalization group (VS-IMSRG) calculations including contributions of two-body currents as well as to shell-model calculations. The two-body-current contributions significantly improve the agreement with experimental data, reducing the mean-square deviation from the experimental moments by a factor of 3 to 5, depending on the employed interaction. For all interactions, the largest contributions are obtained for the $5/2^-$ ($7/2^-$) isotopes $^{65}$Ni ($^{55}$Ni), which is ascribed to the high angular momentum of the $f$ orbitals. Our results demonstrate that the inclusion of two-body-current contributions to the magnetic moment in an isotopic chain of complex nuclei can be handled by the VS-IMSRG and can outperform phenomenological shell-model calculations using effective $g$-factors in the nickel region., Comment: Published in Physics Letters B, 10 pages, 3 figures

Tokamak plasmas with strong negative triangularity (NT) shaping typically exhibit fundamentally different edge behavior than conventional L-mode or H-mode plasmas. Over the entire DIII-D database, plasmas with sufficiently negative triangularity are found to be inherently free of edge localized modes (ELMs), even at injected powers well above the predicted L-H power threshold. A critical triangularly ($\delta_\mathrm{crit}\simeq-0.15$), consistent with inherently ELM-free operation is identified, beyond which access to the second stability region for infinite-$n$ ballooning modes closes on DIII-D. It is also possible to close access to this region, and thereby prevent an H-mode transition, at weaker average triangularities ($\delta\lesssim\delta_\mathrm{crit}$) provided that at least one of the two x-points is still sufficiently negative. Enhanced low field side magnetic fluctuations during ELM-free operation are consistent with additional turbulence limiting the NT edge gradient. Despite the reduced upper limit on the pressure gradient imposed by ballooning stability, NT plasmas are able to support small pedestals and are typically characterized by an enhancement of edge pressure gradients beyond those found in traditional L-mode plasmas. Further, the pressure gradient inside of this small pedestal is unusually steep, allowing access to high core performance that is competitive with other ELM-free regimes previously achieved on DIII-D. Since ELM-free operation in NT is linked directly to the magnetic geometry, NT fusion pilot plants are predicted to maintain advantageous edge conditions even in burning plasma regimes, potentially eliminating reactor core-integration issues caused by ELMs.

Artificial Intelligence applications gradually move outside the safe walls of research labs and invade our daily lives. This is also true for Machine Learning methods on Knowledge Graphs, which has led to a steady increase in their application since the beginning of the 21st century. However, in many applications, users require an explanation of the Artificial Intelligences decision. This led to increased demand for Comprehensible Artificial Intelligence. Knowledge Graphs epitomize fertile soil for Comprehensible Artificial Intelligence, due to their ability to display connected data, i.e. knowledge, in a human- as well as machine-readable way. This survey gives a short history to Comprehensible Artificial Intelligence on Knowledge Graphs. Furthermore, we contribute by arguing that the concept Explainable Artificial Intelligence is overloaded and overlapping with Interpretable Machine Learning. By introducing the parent concept Comprehensible Artificial Intelligence, we provide a clear-cut distinction of both concepts while accounting for their similarities. Thus, we provide in this survey a case for Comprehensible Artificial Intelligence on Knowledge Graphs consisting of Interpretable Machine Learning on Knowledge Graphs and Explainable Artificial Intelligence on Knowledge Graphs. This leads to the introduction of a novel taxonomy for Comprehensible Artificial Intelligence on Knowledge Graphs. In addition, a comprehensive overview of the research on Comprehensible Artificial Intelligence on Knowledge Graphs is presented and put into the context of the taxonomy. Finally, research gaps in the field of Comprehensible Artificial Intelligence on Knowledge Graphs are identified for future research.

In blind quantum computing, a user with a simple client device can perform a quantum computation on a remote quantum server such that the server cannot gain knowledge about the computation. Here, we numerically investigate hardware requirements for verifiable blind quantum computing using an ion trap as server and a distant measurement-only client. While the client has no direct access to quantum-computing resources, it can remotely execute quantum programs on the server by measuring photons emitted by the trapped ion. We introduce a numerical model for trapped-ion quantum devices in NetSquid, a discrete-event simulator for quantum networks. Using this, we determine the minimal hardware requirements on a per-parameter basis to perform the verifiable blind quantum computing protocol. We benchmark these for a five-qubit linear graph state, with which any single-qubit rotation can be performed, where client and server are separated by 50 km. Current state-of-the-art ion traps satisfy the minimal requirements on a per-parameter basis, but all current imperfections combined make it impossible to perform the blind computation securely over 50 km using existing technology. Using a genetic algorithm, we determine the set of hardware parameters that minimises the total improvements required, finding directions along which to improve hardware to reach our threshold error probability that would enable experimental demonstration. In this way, we lay a path for the near-term experimental progress required to realise the implementation of verifiable blind quantum computing over a 50 km distance.

Large Language Models (LLMs) are often trained with safety guards intended to prevent harmful text generation. However, such safety training can be removed by fine-tuning the LLM on harmful datasets. While this emerging threat (harmful fine-tuning attacks) has been characterized by previous work, there is little understanding of how we should proceed in constructing and validating defenses against these attacks especially in the case where defenders would not have control of the fine-tuning process. We introduce a formal framework based on the training budget of an attacker which we call "Immunization" conditions. Using a formal characterisation of the harmful fine-tuning problem, we provide a thorough description of what a successful defense must comprise of and establish a set of guidelines on how rigorous defense research that gives us confidence should proceed., Comment: Published in EMNLP 2024

Learning rewards from human behaviour or feedback is a promising approach to aligning AI systems with human values but fails to consistently extract correct reward functions. Interpretability tools could enable users to understand and evaluate possible flaws in learned reward functions. We propose Counterfactual Trajectory Explanations (CTEs) to interpret reward functions in reinforcement learning by contrasting an original with a counterfactual partial trajectory and the rewards they each receive. We derive six quality criteria for CTEs and propose a novel Monte-Carlo-based algorithm for generating CTEs that optimises these quality criteria. Finally, we measure how informative the generated explanations are to a proxy-human model by training it on CTEs. CTEs are demonstrably informative for the proxy-human model, increasing the similarity between its predictions and the reward function on unseen trajectories. Further, it learns to accurately judge differences in rewards between trajectories and generalises to out-of-distribution examples. Although CTEs do not lead to a perfect understanding of the reward, our method, and more generally the adaptation of XAI methods, are presented as a fruitful approach for interpreting learned reward functions.

Small interconnected quantum processors can collaborate to tackle quantum computational problems that typically demand more capable devices. These linked processors, referred to as quantum nodes, can use shared entangled states to execute nonlocal operations. As a consequence, understanding how to distribute entangled states among nodes is essential for developing hardware and software. We analyze a protocol where entanglement is continuously distributed among nodes that are physically arranged in a regular pattern: a chain, a honeycomb lattice, a square grid, and a triangular lattice. These regular patterns allow for the modular expansion of networks for large-scale distributed quantum computing. Within the distribution protocol, we investigate how nodes can optimize the frequency of attempting entanglement swaps, trading off multiple entangled states shared with neighboring nodes for fewer states shared with non-neighboring nodes. We evaluate the protocol's performance using the virtual neighborhood size -- a metric indicating the number of other nodes with which a given node shares entangled states. Employing numerical methods, we find that nodes must perform more swaps to maximize the virtual neighborhood size when coherence times are short. In a chain network, the virtual neighborhood size's dependence on swap attempt frequency differs for each node based on its distance from the end of the chain. Conversely, all nodes in the square grid exhibit a qualitatively similar dependence of the virtual neighborhood size on the swap frequency., Comment: 8 pages main text with 4 figures; 11 pages appendices with 7 figures; v2: minor main text refinements (more concise introduction with added reference, more complete discussion), typos corrected

Context: Planet-disk interactions play a crucial role in the understanding of planet formation and disk evolution. There are multiple numerical tools available to simulate these interactions, including the often-used FARGO code and its variants. Many of the codes were extended over time to include additional physical processes with a focus on their accurate modeling. Aims: We introduce FargoCPT, an updated version of FARGO incorporating other previous enhancements to the code, to provide a simulation environment tailored to study interactions between stars, planets, and disks, ensuring accurate representation of planet systems, hydrodynamics, and dust dynamics with a focus on usability. Methods: The radiation-hydrodynamics part of FargoCPT uses a second-order upwind scheme in 2D polar coordinates supporting multiple equations of state, radiation transport, heating and cooling, and self-gravity. Shocks are considered using artificial viscosity. Integration of the N-body system is achieved by leveraging the REBOUND code. The dust module utilizes massless tracer particles, adapting to drag laws for the Stokes and Epstein regimes. Moreover, FargoCPT provides mechanisms to simulate accretion onto the stars and planets. Results: The code has been tested in practice by its use in various publications. Additionally, it comes with an automated test suite to test the physics modules. Conclusions: FargoCPT offers a unique set of simulation capabilities within the current landscape of publicly available planet-disk interaction simulation tools. Its structured interface and underlying technical updates are intended to assist researchers in the ongoing exploration of planet formation., Comment: 31 pages, 18 figures; accepted for publication in Astronomy and Astrophysics (A&A)

Root Cause Analysis (RCA) in the manufacturing of electric vehicles is the process of identifying fault causes. Traditionally, the RCA is conducted manually, relying on process expert knowledge. Meanwhile, sensor networks collect significant amounts of data in the manufacturing process. Using this data for RCA makes it more efficient. However, purely data-driven methods like Causal Bayesian Networks have problems scaling to large-scale, real-world manufacturing processes due to the vast amount of potential cause-effect relationships (CERs). Furthermore, purely data-driven methods have the potential to leave out already known CERs or to learn spurious CERs. The paper contributes by proposing an interactive and intelligent RCA tool that combines expert knowledge of an electric vehicle manufacturing process and a data-driven machine learning method. It uses reasoning over a large-scale Knowledge Graph of the manufacturing process while learning a Causal Bayesian Network. In addition, an Interactive User Interface enables a process expert to give feedback to the root cause graph by adding and removing information to the Knowledge Graph. The interactive and intelligent RCA tool reduces the learning time of the Causal Bayesian Network while decreasing the number of spurious CERs. Thus, the interactive and intelligent RCA tool closes the feedback loop between expert and machine learning method.

In this paper, we explore how techniques from soft robotics can help create a new form of robot expression. We present Sprout, a soft expressive robot that conveys its internal states by changing its body shape. Sprout can extend, bend, twist, and expand using fiber-embedded actuators integrated into its construction. These deformations enable Sprout to express its internal states, for example, by expanding to express anger and bending its body sideways to express curiosity. Through two user studies, we investigated how users interpreted Sprout's expressions, their perceptions of Sprout, and their expectations from future iterations of Sprout's design. We argue that the use of soft actuators opens a novel design space for robot expressions to convey internal states, emotions, and intent., Comment: 10 pages, 5 figures

We present a novel pneumatic augmentation to traditional electric motor-actuated legged robot to increase intermittent power density to perform infrequent explosive hopping behaviors. The pneumatic system is composed of a pneumatic pump, a tank, and a pneumatic actuator. The tank is charged up by the pump during regular hopping motion that is created by the electric motors. At any time after reaching a desired air pressure in the tank, a solenoid valve is utilized to rapidly release the air pressure to the pneumatic actuator (piston) which is used in conjunction with the electric motors to perform explosive hopping, increasing maximum hopping height for one or subsequent cycles. We show that, on a custom-designed one-legged hopping robot, without any additional power source and with this novel pneumatic augmentation system, their associated system identification and optimal control, the robot is able to realize highly explosive hopping with power amplification per cycle by a factor of approximately 5.4 times the power of electric motor actuation alone., Comment: 8 pages, 10 figures. Updated version