(Simplified Abstract) To accomplish breakthroughs in dynamic whole-body locomotion, legged robots have to be terrain aware. Terrain-Aware Locomotion (TAL) implies that the robot can perceive the terrain with its sensors, and can take decisions based on this information. This thesis presents TAL strategies both from a proprioceptive and an exteroceptive perspective. The strategies are implemented at the level of locomotion planning, control, and state estimation, and using optimization and learning techniques. The first part is on TAL strategies at the Whole-Body Control (WBC) level. We introduce a passive WBC (pWBC) framework that allows the robot to stabilize and walk over challenging terrain while taking into account the terrain geometry (inclination) and friction properties. The pWBC relies on rigid contact assumptions which makes it suitable only for stiff terrain. As a consequence, we introduce Soft Terrain Adaptation aNd Compliance Estimation (STANCE) which is a soft terrain adaptation algorithm that generalizes beyond rigid terrain. The second part of the thesis focuses on vision-based TAL strategies. We present Vision-Based Terrain-Aware Locomotion (ViTAL) which is an online planning strategy that selects the footholds based on the robot capabilities, and the robot pose that maximizes the chances of the robot succeeding in reaching these footholds. ViTAL relies on a set of robot skills that characterizes the capabilities of the robot and its legs. The skills include the robot's ability to assess the terrain's geometry, avoid leg collisions, and avoid reaching kinematic limits. Our strategies are based on optimization and learning methods and are validated on HyQ and HyQReal in simulation and experiment. We show that with the help of these strategies, we can push dynamic legged robots one step closer to being fully autonomous and terrain aware.
相關內容
To act in the world, robots rely on a representation of salient task aspects: for example, to carry a cup of coffee, a robot must consider movement efficiency and cup orientation in its behaviour. However, if we want robots to act for and with people, their representations must not be just functional but also reflective of what humans care about, i.e. their representations must be aligned with humans'. In this survey, we pose that current reward and imitation learning approaches suffer from representation misalignment, where the robot's learned representation does not capture the human's representation. We suggest that because humans will be the ultimate evaluator of robot performance in the world, it is critical that we explicitly focus our efforts on aligning learned task representations with humans, in addition to learning the downstream task. We advocate that current representation learning approaches in robotics should be studied from the perspective of how well they accomplish the objective of representation alignment. To do so, we mathematically define the problem, identify its key desiderata, and situate current robot learning methods within this formalism. We conclude the survey by suggesting future directions for exploring open challenges.
Video object segmentation (VOS) aims at segmenting a particular object throughout the entire video clip sequence. The state-of-the-art VOS methods have achieved excellent performance (e.g., 90+% J&F) on existing datasets. However, since the target objects in these existing datasets are usually relatively salient, dominant, and isolated, VOS under complex scenes has rarely been studied. To revisit VOS and make it more applicable in the real world, we collect a new VOS dataset called coMplex video Object SEgmentation (MOSE) to study the tracking and segmenting objects in complex environments. MOSE contains 2,149 video clips and 5,200 objects from 36 categories, with 431,725 high-quality object segmentation masks. The most notable feature of MOSE dataset is complex scenes with crowded and occluded objects. The target objects in the videos are commonly occluded by others and disappear in some frames. To analyze the proposed MOSE dataset, we benchmark 18 existing VOS methods under 4 different settings on the proposed MOSE dataset and conduct comprehensive comparisons. The experiments show that current VOS algorithms cannot well perceive objects in complex scenes. For example, under the semi-supervised VOS setting, the highest J&F by existing state-of-the-art VOS methods is only 59.4% on MOSE, much lower than their ~90% J&F performance on DAVIS. The results reveal that although excellent performance has been achieved on existing benchmarks, there are unresolved challenges under complex scenes and more efforts are desired to explore these challenges in the future. The proposed MOSE dataset has been released at //henghuiding.github.io/MOSE.
In many practical fluid dynamics experiments, measuring variables such as velocity and pressure is possible only at a limited number of sensor locations, or for a few two-dimensional planes in the flow. However, knowledge of the full fields is necessary to understand the dynamics of many flows. Deep learning reconstruction of full flow fields from sparse measurements has recently garnered significant research interest, as a way of overcoming this limitation. This task is referred to as the flow reconstruction (FR) task. In the present study, we propose a convolutional autoencoder based neural network model, dubbed FR3D, which enables FR to be carried out for three-dimensional flows around extruded 3D objects with arbitrary cross-sections. An innovative mapping approach, whereby multiple fluid domains are mapped to an annulus, enables FR3D to generalize its performance to objects not encountered during training. We conclusively demonstrate this generalization capability using a dataset composed of 80 training and 20 testing geometries, all randomly generated. We show that the FR3D model reconstructs pressure and velocity components with a few percentage points of error. Additionally, using these predictions, we accurately estimate the Q-criterion fields as well lift and drag forces on the geometries.
This paper presents an inverse kinematic optimization layer (IKOL) for 3D human pose and shape estimation that leverages the strength of both optimization- and regression-based methods within an end-to-end framework. IKOL involves a nonconvex optimization that establishes an implicit mapping from an image's 3D keypoints and body shapes to the relative body-part rotations. The 3D keypoints and the body shapes are the inputs and the relative body-part rotations are the solutions. However, this procedure is implicit and hard to make differentiable. So, to overcome this issue, we designed a Gauss-Newton differentiation (GN-Diff) procedure to differentiate IKOL. GN-Diff iteratively linearizes the nonconvex objective function to obtain Gauss-Newton directions with closed form solutions. Then, an automatic differentiation procedure is directly applied to generate a Jacobian matrix for end-to-end training. Notably, the GN-Diff procedure works fast because it does not rely on a time-consuming implicit differentiation procedure. The twist rotation and shape parameters are learned from the neural networks and, as a result, IKOL has a much lower computational overhead than most existing optimization-based methods. Additionally, compared to existing regression-based methods, IKOL provides a more accurate mesh-image correspondence. This is because it iteratively reduces the distance between the keypoints and also enhances the reliability of the pose structures. Extensive experiments demonstrate the superiority of our proposed framework over a wide range of 3D human pose and shape estimation methods.
Movement is how people interact with and affect their environment. For realistic character animation, it is necessary to synthesize such interactions between virtual characters and their surroundings. Despite recent progress in character animation using machine learning, most systems focus on controlling an agent's movements in fairly simple and homogeneous environments, with limited interactions with other objects. Furthermore, many previous approaches that synthesize human-scene interactions require significant manual labeling of the training data. In contrast, we present a system that uses adversarial imitation learning and reinforcement learning to train physically-simulated characters that perform scene interaction tasks in a natural and life-like manner. Our method learns scene interaction behaviors from large unstructured motion datasets, without manual annotation of the motion data. These scene interactions are learned using an adversarial discriminator that evaluates the realism of a motion within the context of a scene. The key novelty involves conditioning both the discriminator and the policy networks on scene context. We demonstrate the effectiveness of our approach through three challenging scene interaction tasks: carrying, sitting, and lying down, which require coordination of a character's movements in relation to objects in the environment. Our policies learn to seamlessly transition between different behaviors like idling, walking, and sitting. By randomizing the properties of the objects and their placements during training, our method is able to generalize beyond the objects and scenarios depicted in the training dataset, producing natural character-scene interactions for a wide variety of object shapes and placements. The approach takes physics-based character motion generation a step closer to broad applicability.
Online meta-learning has recently emerged as a marriage between batch meta-learning and online learning, for achieving the capability of quick adaptation on new tasks in a lifelong manner. However, most existing approaches focus on the restrictive setting where the distribution of the online tasks remains fixed with known task boundaries. In this work, we relax these assumptions and propose a novel algorithm for task-agnostic online meta-learning in non-stationary environments. More specifically, we first propose two simple but effective detection mechanisms of task switches and distribution shift based on empirical observations, which serve as a key building block for more elegant online model updates in our algorithm: the task switch detection mechanism allows reusing of the best model available for the current task at hand, and the distribution shift detection mechanism differentiates the meta model update in order to preserve the knowledge for in-distribution tasks and quickly learn the new knowledge for out-of-distribution tasks. In particular, our online meta model updates are based only on the current data, which eliminates the need of storing previous data as required in most existing methods. We further show that a sublinear task-averaged regret can be achieved for our algorithm under mild conditions. Empirical studies on three different benchmarks clearly demonstrate the significant advantage of our algorithm over related baseline approaches.
This paper presents a mobile supernumerary robotic approach to physical assistance in human-robot conjoined actions. The study starts with a description of the SUPER-MAN concept. The idea is to develop and utilize mobile collaborative systems that can follow human loco-manipulation commands to perform industrial tasks through three main components: i) an admittance-type interface, ii) a human-robot interaction controller, and iii) a supernumerary robotic body. Next, we present two possible implementations within the framework from theoretical and hardware perspectives. The first system is called MOCA-MAN and comprises a redundant torque-controlled robotic arm and an omnidirectional mobile platform. The second one is called Kairos-MAN, formed by a high-payload 6-DoF velocity-controlled robotic arm and an omnidirectional mobile platform. The systems share the same admittance interface, through which user wrenches are translated to loco-manipulation commands generated by whole-body controllers of each system. Besides, a thorough user study with multiple and cross-gender subjects is presented to reveal the quantitative performance of the two systems in effort-demanding and dexterous tasks. Moreover, we provide qualitative results from the NASA-TLX questionnaire to demonstrate the SUPER-MAN approach's potential and its acceptability from the users' viewpoint.
Various methods for Multi-Agent Reinforcement Learning (MARL) have been developed with the assumption that agents' policies are based on accurate state information. However, policies learned through Deep Reinforcement Learning (DRL) are susceptible to adversarial state perturbation attacks. In this work, we propose a State-Adversarial Markov Game (SAMG) and make the first attempt to investigate the fundamental properties of MARL under state uncertainties. Our analysis shows that the commonly used solution concepts of optimal agent policy and robust Nash equilibrium do not always exist in SAMGs. To circumvent this difficulty, we consider a new solution concept called robust agent policy, where agents aim to maximize the worst-case expected state value. We prove the existence of robust agent policy for finite state and finite action SAMGs. Additionally, we propose a Robust Multi-Agent Adversarial Actor-Critic (RMA3C) algorithm to learn robust policies for MARL agents under state uncertainties. Our experiments demonstrate that our algorithm outperforms existing methods when faced with state perturbations and greatly improves the robustness of MARL policies. Our code is public on //songyanghan.github.io/what_is_solution/.
Safety in the automotive domain is a well-known topic, which has been in constant development in the past years. The complexity of new systems that add more advanced components in each function has opened new trends that have to be covered from the safety perspective. In this case, not only specifications and requirements have to be covered but also scenarios, which cover all relevant information of the vehicle environment. Many of them are not yet still sufficient defined or considered. In this context, Safety of the Intended Functionality (SOTIF) appears to ensure the system when it might fail because of technological shortcomings or misuses by users. An identification of the plausibly insufficiencies of ADAS/ADS functions has to be done to discover the potential triggering conditions that can lead to these unknown scenarios, which might effect a hazardous behaviour. The main goal of this publication is the definition of an use case to identify these triggering conditions that have been applied to the collision avoidance function implemented in our self-developed mobile Hardware-in-Loop (HiL) platform.
We propose the idea of transferring common-sense knowledge from source categories to target categories for scalable object detection. In our setting, the training data for the source categories have bounding box annotations, while those for the target categories only have image-level annotations. Current state-of-the-art approaches focus on image-level visual or semantic similarity to adapt a detector trained on the source categories to the new target categories. In contrast, our key idea is to (i) use similarity not at image-level, but rather at region-level, as well as (ii) leverage richer common-sense (based on attribute, spatial, etc.,) to guide the algorithm towards learning the correct detections. We acquire such common-sense cues automatically from readily-available knowledge bases without any extra human effort. On the challenging MS COCO dataset, we find that using common-sense knowledge substantially improves detection performance over existing transfer-learning baselines.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191