Search

In visual Reinforcement Learning (RL), learning from pixel-based observations poses significant challenges on sample efficiency, primarily due to the complexity of extracting informative state representations from high-dimensional data. Previous methods such as contrastive-based approaches have made strides in improving sample efficiency but fall short in modeling the nuanced evolution of states. To address this, we introduce MOOSS, a novel framework that leverages a temporal contrastive objective with the help of graph-based spatial-temporal masking to explicitly model state evolution in visual RL. Specifically, we propose a self-supervised dual-component strategy that integrates (1) a graph construction of pixel-based observations for spatial-temporal masking, coupled with (2) a multi-level contrastive learning mechanism that enriches state representations by emphasizing temporal continuity and change of states. MOOSS advances the understanding of state dynamics by disrupting and learning from spatial-temporal correlations, which facilitates policy learning. Our comprehensive evaluation on multiple continuous and discrete control benchmarks shows that MOOSS outperforms previous state-of-the-art visual RL methods in terms of sample efficiency, demonstrating the effectiveness of our method. Our code is released at https://github.com/jsun57/MOOSS., Comment: WACV 2025

Under-canopy agricultural robots require robust navigation capabilities to enable full autonomy but struggle with tight row turning between crop rows due to degraded GPS reception, visual aliasing, occlusion, and complex vehicle dynamics. We propose an imitation learning approach using diffusion policies to learn row turning behaviors from demonstrations provided by human operators or privileged controllers. Simulation experiments in a corn field environment show potential in learning this task with only visual observations and velocity states. However, challenges remain in maintaining control within rows and handling varied initial conditions, highlighting areas for future improvement., Comment: Accepted as Extended Abstract to the IEEE ICRA@40 2024

Tracking plant features is crucial for various agricultural tasks like phenotyping, pruning, or harvesting, but the unstructured, cluttered, and deformable nature of plant environments makes it a challenging task. In this context, the recent advancements in foundational models show promise in addressing this challenge. In our work, we propose PlantTrack where we utilize DINOv2 which provides high-dimensional features, and train a keypoint heatmap predictor network to identify the locations of semantic features such as fruits and leaves which are then used as prompts for point tracking across video frames using TAPIR. We show that with as few as 20 synthetic images for training the keypoint predictor, we achieve zero-shot Sim2Real transfer, enabling effective tracking of plant features in real environments.

Teams are the fundamental units propelling innovation and advancing modern science. A rich literature links the fundamental features of teams, such as their size and diversity, to academic success. However, such analyses fail to capture temporal patterns, treating each group of co-authors as a distinct unit and neglecting the existence of persistent collaborations. By contrast, teams are dynamical entities, made of core members who consistently work together, surrounded by transient members who sporadically participate. Leveraging on a large dataset of over 205 million scientific papers published since 1900, we extract 511,550 core teams of statistically significant persistent collaborations of pairs and larger groups of scientists. We look into `team careers' investigating their trajectories in time, characterizing their formation, productivity and eventual dissolution. We characterize team composition along multiple dimensions, including age, academic affiliation and scientific disciplines. Finally, we investigate the academic impact of persistent collaborations, hallmarking the key compositional features underlying their success. Our work sheds light on the nature of persistent teams, informing researchers, institutions and funding agencies about the dynamics of their formation, evolution and success., Comment: 18 pages, 13 figures

We study fairness in social influence maximization, whereby one seeks to select seeds that spread a given information throughout a network, ensuring balanced outreach among different communities (e.g. demographic groups). In the literature, fairness is often quantified in terms of the expected outreach within individual communities. In this paper, we demonstrate that such fairness metrics can be misleading since they ignore the stochastic nature of information diffusion processes. When information diffusion occurs in a probabilistic manner, multiple outreach scenarios can occur. As such, outcomes such as "in 50% of the cases, no one of group 1 receives the information and everyone in group 2 receives it and in other 50%, the opposite happens", which always results in largely unfair outcomes, are classified as fair by a variety of fairness metrics in the literature. We tackle this problem by designing a new fairness metric, mutual fairness, that captures variability in outreach through optimal transport theory. We propose a new seed selection algorithm that optimizes both outreach and mutual fairness, and we show its efficacy on several real datasets. We find that our algorithm increases fairness with only a minor decrease (and at times, even an increase) in efficiency.

Successful deployment of mobile robots in unstructured domains requires an understanding of the environment and terrain to avoid hazardous areas, getting stuck, and colliding with obstacles. Traversability estimation--which predicts where in the environment a robot can travel--is one prominent approach that tackles this problem. Existing geometric methods may ignore important semantic considerations, while semantic segmentation approaches involve a tedious labeling process. Recent self-supervised methods reduce labeling tedium, but require additional data or models and tend to struggle to explicitly label untraversable areas. To address these limitations, we introduce a weakly-supervised method for relative traversability estimation. Our method involves manually annotating the relative traversability of a small number of point pairs, which significantly reduces labeling effort compared to traditional segmentation-based methods and avoids the limitations of self-supervised methods. We further improve the performance of our method through a novel cross-image labeling strategy and loss function. We demonstrate the viability and performance of our method through deployment on a mobile robot in outdoor environments., Comment: Accepted by RA-L. Code is available at https://github.com/andreschreiber/W-RIZZ

As the demand for food surges and the agricultural sector undergoes a transformative shift towards sustainability and efficiency, the need for precise and proactive measures to ensure the health and welfare of livestock becomes paramount. In the context of the broader agricultural landscape outlined, the application of Hyperspectral Imaging (HSI) takes on profound significance. HSI has emerged as a cutting-edge, non-destructive technique for fast and accurate egg quality analysis, including the detection of chick embryo mortality. However, the high cost and operational complexity compared to conventional RGB imaging are significant bottlenecks in the widespread adoption of HSI technology. To overcome these hurdles and unlock the full potential of HSI, a promising solution is hyperspectral image reconstruction from standard RGB images. This study aims to reconstruct hyperspectral images from RGB images for non-destructive early prediction of chick embryo mortality. Firstly, the performance of different image reconstruction algorithms, such as HRNET, MST++, Restormer, and EDSR were compared to reconstruct the hyperspectral images of the eggs in the early incubation period. Later, the reconstructed spectra were used to differentiate live from dead chick-producing eggs using the XGBoost and Random Forest classification methods. Among the reconstruction methods, HRNET showed impressive reconstruction performance with MRAE of 0.0955, RMSE of 0.0159, and PSNR of 36.79 dB. This study motivated that harnessing imaging technology integrated with smart sensors and data analytics has the potential to improve automation, enhance biosecurity, and optimize resource management towards sustainable agriculture 4.0., Comment: Under review

We present a vision-based navigation system for under-canopy agricultural robots using semantic keypoints. Autonomous under-canopy navigation is challenging due to the tight spacing between the crop rows ($\sim 0.75$ m), degradation in RTK-GPS accuracy due to multipath error, and noise in LiDAR measurements from the excessive clutter. Our system, CropFollow++, introduces modular and interpretable perception architecture with a learned semantic keypoint representation. We deployed CropFollow++ in multiple under-canopy cover crop planting robots on a large scale (25 km in total) in various field conditions and we discuss the key lessons learned from this., Comment: Accepted to the IEEE ICRA Workshop on Field Robotics 2024

The parallel and distributed processing are becoming de facto industry standard, and a large part of the current research is targeted on how to make computing scalable and distributed, dynamically, without allocating the resources on permanent basis. The present article focuses on the study and performance of distributed and parallel algorithms their file systems, to achieve scalability at local level (OpenMP platform), and at global level where computing and file systems are distributed. Various applications, algorithms,file systems have been used to demonstrate the areas, and their performance studies have been presented. The systems and applications chosen here are of open-source nature, due to their wider applicability., Comment: 8 pages, 5 figures

Open-world video instance segmentation is an important video understanding task. Yet most methods either operate in a closed-world setting, require an additional user-input, or use classic region-based proposals to identify never before seen objects. Further, these methods only assign a one-word label to detected objects, and don't generate rich object-centric descriptions. They also often suffer from highly overlapping predictions. To address these issues, we propose Open-World Video Instance Segmentation and Captioning (OW-VISCap), an approach to jointly segment, track, and caption previously seen or unseen objects in a video. For this, we introduce open-world object queries to discover never before seen objects without additional user-input. We generate rich and descriptive object-centric captions for each detected object via a masked attention augmented LLM input. We introduce an inter-query contrastive loss to ensure that the object queries differ from one another. Our generalized approach matches or surpasses state-of-the-art on three tasks: open-world video instance segmentation on the BURST dataset, dense video object captioning on the VidSTG dataset, and closed-world video instance segmentation on the OVIS dataset., Comment: Project page: https://anwesachoudhuri.github.io/OpenWorldVISCap/

Large-scale imaging surveys have grown about 1000 times faster than the number of astronomers in the last 3 decades. Using Artificial Intelligence instead of astronomer's brains for interpretative tasks allows astronomers to keep up with the data. We give a progress report on using Convolutional Neural Networks (CNNs) to classify three classes of rare objects (galaxy mergers, strong gravitational lenses and asteroids) in the Kilo-Degree Survey (KiDS) and the Euclid Survey., Comment: ADASS XXXII - 2022, Victoria, Conference Proceedings

Neural Machine Translation (NMT) continues to improve in quality and adoption, yet the inadvertent perpetuation of gender bias remains a significant concern. Despite numerous studies on gender bias in translations into English from weakly gendered-languages, there are no benchmarks for evaluating this phenomenon or for assessing mitigation strategies. To address this gap, we introduce GATE X-E, an extension to the GATE (Rarrick et al., 2023) corpus, that consists of human translations from Turkish, Hungarian, Finnish, and Persian into English. Each translation is accompanied by feminine, masculine, and neutral variants. The dataset, which contains between 1250 and 1850 instances for each of the four language pairs, features natural sentences with a wide range of sentence lengths and domains, challenging translation rewriters on various linguistic phenomena. Additionally, we present a translation gender rewriting solution built with GPT-4 and use GATE X-E to evaluate it. We open source our contributions to encourage further research on gender debiasing., Comment: arXiv admin note: substantial text overlap with arXiv:2311.08836

A vital aspect of Indian Classical Music (ICM) is Raga, which serves as a melodic framework for compositions and improvisations alike. Raga Recognition is an important music information retrieval task in ICM as it can aid numerous downstream applications ranging from music recommendations to organizing huge music collections. In this work, we propose a deep learning based approach to Raga recognition. Our approach employs efficient pre possessing and learns temporal sequences in music data using Long Short Term Memory based Recurrent Neural Networks (LSTM-RNN). We train and test the network on smaller sequences sampled from the original audio while the final inference is performed on the audio as a whole. Our method achieves an accuracy of 88.1% and 97 % during inference on the Comp Music Carnatic dataset and its 10 Raga subset respectively making it the state-of-the-art for the Raga recognition task. Our approach also enables sequence ranking which aids us in retrieving melodic patterns from a given music data base that are closely related to the presented query sequence.

Robotic harvesting has the potential to positively impact agricultural productivity, reduce costs, improve food quality, enhance sustainability, and to address labor shortage. In the rapidly advancing field of agricultural robotics, the necessity of training robots in a virtual environment has become essential. Generating training data to automatize the underlying computer vision tasks such as image segmentation, object detection and classification, also heavily relies on such virtual environments as synthetic data is often required to overcome the shortage and lack of variety of real data sets. However, physics engines commonly employed within the robotics community, such as ODE, Simbody, Bullet, and DART, primarily support motion and collision interaction of rigid bodies. This inherent limitation hinders experimentation and progress in handling non-rigid objects such as plants and crops. In this contribution, we present a plugin for the Gazebo simulation platform based on Cosserat rods to model plant motion. It enables the simulation of plants and their interaction with the environment. We demonstrate that, using our plugin, users can conduct harvesting simulations in Gazebo by simulating a robotic arm picking fruits and achieve results comparable to real-world experiments., Comment: Upon request, we are happy to share our GazeboPlants plugin open-source (MPL 2.0)

Accurate and robust navigation in unstructured environments requires fusing data from multiple sensors. Such fusion ensures that the robot is better aware of its surroundings, including areas of the environment that are not immediately visible but were visible at a different time. To solve this problem, we propose a method for traversability prediction in challenging outdoor environments using a sequence of RGB and depth images fused with pose estimations. Our method, termed WayFASTER (Waypoints-Free Autonomous System for Traversability with Enhanced Robustness), uses experience data recorded from a receding horizon estimator to train a self-supervised neural network for traversability prediction, eliminating the need for heuristics. Our experiments demonstrate that our method excels at avoiding obstacles, and correctly detects that traversable terrains, such as tall grass, can be navigable. By using a sequence of images, WayFASTER significantly enhances the robot's awareness of its surroundings, enabling it to predict the traversability of terrains that are not immediately visible. This enhanced awareness contributes to better navigation performance in environments where such predictive capabilities are essential., Comment: Accepted for the 2024 IEEE International Conference on Robotics and Automation (ICRA 2024)

Advancements in technology have recently allowed us to collect and analyse large-scale fine-grained data about human performance, drastically changing the way we approach sports. Here, we provide the first comprehensive analysis of individual and team performance in One-Day International cricket, one of the most popular sports in the world. We investigate temporal patterns of individual success by quantifying the location of the best performance of a player and find that they can happen at any time in their career, surrounded by a burst of comparable top performances. Our analysis shows that long-term performance can be predicted from early observations and that temporary exclusions of players from teams are often due to declining performances but are also associated with strong comebacks. By computing the duration of streaks of winning performances compared to random expectations, we demonstrate that teams win and lose matches consecutively. We define the contributions of specialists such as openers, all-rounders and wicket-keepers and show that a balanced performance from multiple individuals is required to ensure team success. Finally, we measure how transitioning to captaincy in the team improves the performance of batsmen, but not that of bowlers. Our work emphasizes how individual endeavours and team dynamics interconnect and influence collective outcomes in sports., Comment: 11 pages, 5 figures in main. Supplemental material with 9 figures

Machine Translation (MT) continues to improve in quality and adoption, yet the inadvertent perpetuation of gender bias remains a significant concern. Despite numerous studies into gender bias in translations from gender-neutral languages such as Turkish into more strongly gendered languages like English, there are no benchmarks for evaluating this phenomenon or for assessing mitigation strategies. To address this gap, we introduce GATE X-E, an extension to the GATE (Rarrick et al., 2023) corpus, that consists of human translations from Turkish, Hungarian, Finnish, and Persian into English. Each translation is accompanied by feminine, masculine, and neutral variants for each possible gender interpretation. The dataset, which contains between 1250 and 1850 instances for each of the four language pairs, features natural sentences with a wide range of sentence lengths and domains, challenging translation rewriters on various linguistic phenomena. Additionally, we present an English gender rewriting solution built on GPT-3.5 Turbo and use GATE X-E to evaluate it. We open source our contributions to encourage further research on gender debiasing.

Convolutional neural networks (CNNs) are the state-of-the-art technique for identifying strong gravitational lenses. Although they are highly successful in recovering genuine lens systems with a high true-positive rate, the unbalanced nature of the data set (lens systems are rare), still leads to a high false positive rate. For these techniques to be successful in upcoming surveys (e.g. with Euclid) most emphasis should be set on reducing false positives, rather than on reducing false negatives. In this paper, we introduce densely connected neural networks (DenseNets) as the CNN architecture in a new pipeline-ensemble model containing an ensemble of classification CNNs and regression CNNs to classify and rank-order lenses, respectively. We show that DenseNets achieve comparable true positive rates but considerably lower false positive rates (when compared to residual networks; ResNets). Thus, we recommend DenseNets for future missions involving large data sets, such as Euclid, where low false positive rates play a key role in the automated follow-up and analysis of large numbers of strong gravitational lens candidates when human vetting is no longer feasible

Arrival of multicore systems has enforced a new scenario in computing, the parallel and distributed algorithms are fast replacing the older sequential algorithms, with many challenges of these techniques. The distributed algorithms provide distributed processing using distributed file systems and processing units, while network is modeled as minimum cost spanning tree. On the other hand, the parallel processing chooses different language platforms, data parallel vs. parallel programming, and GPUs. Processing units, memory elements and storage are connected through dynamic distributed networks in the form of spanning trees. The article presents foundational algorithms, analysis, and efficiency considerations., Comment: 9 pages

In the recent progress in embodied navigation and sim-to-robot transfer, modular policies have emerged as a de facto framework. However, there is more to compositionality beyond the decomposition of the learning load into modular components. In this work, we investigate a principled way to syntactically combine these components. Particularly, we propose Exploitation-Guided Exploration (XGX) where separate modules for exploration and exploitation come together in a novel and intuitive manner. We configure the exploitation module to take over in the deterministic final steps of navigation i.e. when the goal becomes visible. Crucially, an exploitation module teacher-forces the exploration module and continues driving an overridden policy optimization. XGX, with effective decomposition and novel guidance, improves the state-of-the-art performance on the challenging object navigation task from 70% to 73%. Along with better accuracy, through targeted analysis, we show that XGX is also more efficient at goal-conditioned exploration. Finally, we show sim-to-real transfer to robot hardware and XGX performs over two-fold better than the best baseline from simulation benchmarking. Project page: xgxvisnav.github.io, Comment: Code and results available at http://xgxvisnav.github.io

We study the sensitivity of infinite-dimensional Bayesian linear inverse problems governed by partial differential equations (PDEs) with respect to modeling uncertainties. In particular, we consider derivative-based sensitivity analysis of the information gain, as measured by the Kullback-Leibler divergence from the posterior to the prior distribution. To facilitate this, we develop a fast and accurate method for computing derivatives of the information gain with respect to auxiliary model parameters. Our approach combines low-rank approximations, adjoint-based eigenvalue sensitivity analysis, and post-optimal sensitivity analysis. The proposed approach also paves way for global sensitivity analysis by computing derivative-based global sensitivity measures. We illustrate different aspects of the proposed approach using an inverse problem governed by a scalar linear elliptic PDE, and an inverse problem governed by the three-dimensional equations of linear elasticity, which is motivated by the inversion of the fault-slip field after an earthquake., Comment: 20 pages, 7 figures

Roll solutions at the zigzag boundary, typically selected by patterns and defects in numerical simulations, are shown to be nonlinearly stable. This result also serves as an example that linear decay weaker than the classical diffusive decay, together with quadratic nonlinearity, still gives nonlinear stability of spatially periodic patterns. Linear analysis reveals that, instead of the classical $t^{-1}$ diffusive decay rate, small perturbations of roll solutions at the zigzag boundary decay with a $t^{-3/4}$ rate along with time, due to the degeneracy of the quadratic term of the continuation of the translational mode of the linearized operator in the Bloch-Fourier spaces. The nonlinear stability proof is based on a decomposition of the neutral translational mode and the faster decaying modes in the Bloch-Fourier space, and a fixed-point argument, demonstrating the irrelevancy of the nonlinear terms., Comment: 54 pages, 1 figure

Traffic forecasting, crucial for urban planning, requires accurate predictions of spatial-temporal traffic patterns across urban areas. Existing research mainly focuses on designing complex models that capture spatial-temporal dependencies among variables explicitly. However, this field faces challenges related to data scarcity and model stability, which results in limited performance improvement. To address these issues, we propose Spatial-Temporal Masked AutoEncoders (STMAE), a plug-and-play framework designed to enhance existing spatial-temporal models on traffic prediction. STMAE consists of two learning stages. In the pretraining stage, an encoder processes partially visible traffic data produced by a dual-masking strategy, including biased random walk-based spatial masking and patch-based temporal masking. Subsequently, two decoders aim to reconstruct the masked counterparts from both spatial and temporal perspectives. The fine-tuning stage retains the pretrained encoder and integrates it with decoders from existing backbones to improve forecasting accuracy. Our results on traffic benchmarks show that STMAE can largely enhance the forecasting capabilities of various spatial-temporal models., Comment: Accepted by CIKM 2024

Global Climate Model (GCM) tuning (calibration) is a tedious and time-consuming process, with high-dimensional input and output fields. Experts typically tune by iteratively running climate simulations with hand-picked values of tuning parameters. Many, in both the statistical and climate literature, have proposed alternative calibration methods, but most are impractical or difficult to implement. We present a practical, robust and rigorous calibration approach on the atmosphere-only model of the Department of Energy's Energy Exascale Earth System Model (E3SM) version 2. Our approach can be summarized into two main parts: (1) the training of a surrogate that predicts E3SM output in a fraction of the time compared to running E3SM, and (2) gradient-based parameter optimization. To train the surrogate, we generate a set of designed ensemble runs that span our input parameter space and use polynomial chaos expansions on a reduced output space to fit the E3SM output. We use this surrogate in an optimization scheme to identify values of the input parameters for which our model best matches gridded spatial fields of climate observations. To validate our choice of parameters, we run E3SMv2 with the optimal parameter values and compare prediction results to expertly-tuned simulations across 45 different output fields. This flexible, robust, and automated approach is straightforward to implement, and we demonstrate that the resulting model output matches present day climate observations as well or better than the corresponding output from expert tuned parameter values, while considering high-dimensional output and operating in a fraction of the time.

Action potential waves triggering the heart's contractions can be imaged at high spatial and temporal resolutions across the heart surface using voltage-sensitive optical mapping. However, for over three decades, optical mapping has been performed with contraction-inhibited hearts. While it was recently demonstrated that action potential waves can be imaged on parts of the three-dimensional deforming ventricular surface using multi-camera optical mapping, panoramic measurements of action potential waves across the entire beating heart surface remained elusive. Here, we introduce a high-resolution multi-camera optical mapping system consisting of up to 24 high-speed, low-cost cameras with which it is possible to image action potential waves at high resolutions on the entire, strongly deforming ventricular surface of the heart. We imaged isolated hearts inside a custom-designed soccerball-shaped imaging chamber, which facilitates imaging and even illumination with excitation light from all sides in a panoramic fashion. We found that it is possible to image the entire ventricular surface using 12 cameras with 0.5-1.0 megapixels combined resolution. The 12 calibrated cameras generate 1.5 gigabytes of video data per second at imaging speeds of 500 fps, which we process using various computer vision techniques, including three-dimensional cooperative multi-view motion tracking, to generate three-dimensional dynamic reconstructions of the deforming heart surface with corresponding high-resolution voltage-sensitive optical measurements. With our setup, we measured action potential waves at unprecedented resolutions on the contracting three-dimensional surface of rabbit hearts during sinus rhythm, paced rhythm as well as ventricular fibrillation. Our imaging setup defines a new state-of-the-art in the field and can be used to study the heart's electromechanical dynamics during health and disease., Comment: Cardiovascular Research, Biophysics, Biomedical Engineering, Optics, Physics, Computer Vision

Underwater communication signals typically suffer from distortion due to motion-induced Doppler. Especially in shallow water environments, recovering the signal is challenging due to the time-varying Doppler effects distorting each path differently. However, conventional Doppler estimation algorithms typically model uniform Doppler across all paths and often fail to provide robust Doppler tracking in multipath environments. In this paper, we propose a dynamic programming-inspired method, called online segmented recursive least-squares (OSRLS) to sequentially estimate the time-varying non-uniform Doppler across different multipath arrivals. By approximating the non-linear time distortion as a piece-wise-linear Markov model, we formulate the problem in a dynamic programming framework known as segmented least-squares (SLS). In order to circumvent an ill-conditioned formulation, perturbations are added to the Doppler model during the linearization process. The successful operation of the algorithm is demonstrated in a simulation on a synthetic channel with time-varying non-uniform Doppler.

Stochastic Human Motion Prediction (HMP) aims to predict multiple possible future human pose sequences from observed ones. Most prior works learn motion distributions through encoding-decoding in the latent space, which does not preserve motion's spatial-temporal structure. While effective, these methods often require complex, multi-stage training and yield predictions that are inconsistent with the provided history and can be physically unrealistic. To address these issues, we propose CoMusion, a single-stage, end-to-end diffusion-based stochastic HMP framework. CoMusion is inspired from the insight that a smooth future pose initialization improves prediction performance, a strategy not previously utilized in stochastic models but evidenced in deterministic works. To generate such initialization, CoMusion's motion predictor starts with a Transformer-based network for initial reconstruction of corrupted motion. Then, a graph convolutional network (GCN) is employed to refine the prediction considering past observations in the discrete cosine transformation (DCT) space. Our method, facilitated by the Transformer-GCN module design and a proposed variance scheduler, excels in predicting accurate, realistic, and consistent motions, while maintaining appropriate diversity. Experimental results on benchmark datasets demonstrate that CoMusion surpasses prior methods across metrics, while demonstrating superior generation quality. Our Code is released at https://github.com/jsun57/CoMusion/ ., Comment: ECCV 2024

Human motion prediction aims to forecast an upcoming pose sequence given a past human motion trajectory. To address the problem, in this work we propose FreqMRN, a human motion prediction framework that takes into account both the kinematic structure of the human body and the temporal smoothness nature of motion. Specifically, FreqMRN first generates a fixed-size motion history summary using a motion attention module, which helps avoid inaccurate motion predictions due to excessively long motion inputs. Then, supervised by the proposed spatial-temporal-aware, velocity-aware and global-smoothness-aware losses, FreqMRN iteratively refines the predicted motion though the proposed motion refinement module, which converts motion representations back and forth between pose space and frequency space. We evaluate FreqMRN on several standard benchmark datasets, including Human3.6M, AMASS and 3DPW. Experimental results demonstrate that FreqMRN outperforms previous methods by large margins for both short-term and long-term predictions, while demonstrating superior robustness.

This paper presents a deep learning based model predictive control (MPC) algorithm for systems with unmatched and bounded state-action dependent uncertainties of unknown structure. We utilize a deep neural network (DNN) as an oracle in the underlying optimization problem of learning based MPC (LBMPC) to estimate unmatched uncertainties. Generally, non-parametric oracles such as DNN are considered difficult to employ with LBMPC due to the technical difficulties associated with estimation of their coefficients in real time. We employ a dual-timescale adaptation mechanism, where the weights of the last layer of the neural network are updated in real time while the inner layers are trained on a slower timescale using the training data collected online and selectively stored in a buffer. Our results are validated through a numerical experiment on the compression system model of jet engine. These results indicate that the proposed approach is implementable in real time and carries the theoretical guarantees of LBMPC.