ICRA 2021 论文列表

The lack of stability guarantee restricts the practical use of learning-based methods in core control problems in robotics. We develop new methods for learning neural control policies and neural Lyapunov critic functions in the modelfree reinforcement learning (RL) setting. We use sample-based approaches and the Almost Lyapunov function conditions to estimate the region of attraction and invariance properties through the learned Lyapunov critic functions. The methods enhance stability of neural controllers for various nonlinear systems including automobile and quadrotor control.

Soft optical tactile sensors enable robots to manipulate deformable objects by capturing important features such as high-resolution contact geometry and estimations of object compliance. This work presents a variable stiffness soft tactile end-effector called StRETcH, a Soft to Resistive Elastic Tactile Hand, that is easily manufactured and integrated with a robotic arm. An elastic membrane is suspended between two robotic fingers, and a depth sensor capturing the deformations of the elastic membrane enables sub-millimeter accurate estimates of contact geometries. The parallel-jaw gripper varies the stiffness of the membrane by uniaxially stretching it, which controllably modulates StRETcH’s effective modulus from approximately 4kPa to 9kPa. This work uses StRETcH to reconstruct the contact geometry of rigid and deformable objects, estimate the stiffness of four balloons filled with different substances, and manipulate dough into a desired shape.

This paper presents the design and preliminary evaluation of a portable spring-based knee exoskeleton, the SpringExo, which is designed to provide assistance to the leg while minimizing interference with the natural leg movement. Traditional rigid exoskeletons are unable to accurately align with a user’s anatomical joints. In addition, the user’s natural motion pattern is likely to change due to the constraints of the rigid exoskeleton. Though some textile-based soft exosuits and cable-driven soft exoskeletons have been developed to achieve better alignment with human’s biological joints, forces applied by the cables have to be sustained by human skeleton and joints. SpringExo, in comparison, uses a coil spring which the user wears around the thigh and shank, and does not require alignment with the joints. The spring stores energy and provides minimal interference during elastic deformation. A key feature of the SpringExo is that the springs store energy during the flexion phase and release this energy to assist the knee extension in the extension phase. We conducted human subjects study to verify its biomechanical and physiological effects on the user during stair ascent. Results from a six-subject study showed that the device did not interfere with the natural joint angles and assisted knee extension during stair ascent. However, further redesign and optimization are needed on the actuation system to offset SpringExo’s drawback of hindering knee flexion.

We address the problem of autonomous exploration of unknown environments using a Micro Aerial Vehicle (MAV) equipped with an active depth sensor. As such, the task consists in mapping the gradually discovered environment while planning the envisioned trajectories in real-time, using on-board computation only. To do so, we present SplatPlanner, an end-to-end autonomous planner that is based on a novel Permutohedral Frontier Filtering (PFF) which relies on a combination of highly efficient operations stemming from bilateral filtering using permutohedral lattices to guide the entire exploration. In particular, our PFF is computationally linear in input size, nearly parameter-free, and aggregates spatial information about frontier-neighborhoods into density scores in one single step. Comparative experiments made on simulated environments of increasing complexity show our method consistently outperforms recent state-of-the-art methods in terms of computational efficiency, exploration speed and qualitative coverage of scenes. Finally, we also display the practical capabilities of our end-to-end system in a challenging real-flight scenario.

In this paper, we propose a robust end-to-end multi-modal pipeline for place recognition where the sensor systems can differ from the map building to the query. Our approach operates directly on images and LiDAR scans without requiring any local feature extraction modules. By projecting the sensor data onto the unit sphere, we learn a multi-modal descriptor of partially overlapping scenes using a spherical convolutional neural network. The employed spherical projection model enables the support of arbitrary LiDAR and camera systems readily without losing information. Loop closure candidates are found using a nearest-neighbor lookup in the embedding space. We tackle the problem of correctly identifying the closest place by correlating the candidates’ power spectra, obtaining a confidence value per prospect. Our estimate for the correct place corresponds then to the candidate with the highest confidence. We evaluate our proposal w.r.t. state-of-the-art approaches in place recognition using real-world data acquired using different sensors. Our approach can achieve a recall that is up to 10% and 5% higher than for a LiDAR- and vision-based system, respectively, when the sensor setup differs between model training and deployment. Additionally, our place selection can correctly identify up to 95% matches from the candidate set.

In this paper, we present a spherical magnetic joint for the inverted locomotion of a multi-legged robot. The permanent magnet’s spherical shape allows the robot to attach its foot to a steel surface without energy consumption. However, the robot’s inverted locomotion requires foot flexibility for placement and gait construction of the robot. Therefore, the spherical magnetic joint mechanism was designed and implemented for the robot feet to deal with angular placement. For decoupling the foot from the steel surface, the attractive force is adjusted by tilting the adjustable sleeve mechanism at an adequate angle between the surface and foot tip. Experimental results show that the spherical magnetic joint can maintain the attractive force at any angle, and the sleeve mechanism can reduce 20% of the reaction force for pulling the legs from the steel surfaces. Furthermore, the designed gait for inverted locomotion with a spherical magnetic joint was tested and compared to prove the concept of the spherical magnetic joint and sleeve mechanism.

An effective understanding of the contextual environment and accurate motion forecasting of surrounding agents is crucial for the development of autonomous vehicles and social mobile robots. This task is challenging since the behavior of an autonomous agent is not only affected by its own intention, but also by the static environment and surrounding dynamically interacting agents. Previous works focused on utilizing the spatial and temporal information in time domain while not sufficiently taking advantage of the cues in frequency domain. To this end, we propose a Spectral Temporal Graph Neural Network (SpecTGNN), which can capture inter-agent correlations and temporal dependency simultaneously in frequency domain in addition to time domain. SpecTGNN operates on both an agent graph with dynamic state information and an environment graph with the features extracted from context images in two streams. The model integrates graph Fourier transform, spectral graph convolution and temporal gated convolution to encode history information and forecast future trajectories. Moreover, we incorporate a multi-head spatio-temporal attention mechanism to mitigate the effect of error propagation in a long time horizon. We demonstrate the performance of SpecTGNN on two public trajectory prediction benchmark datasets, which achieves state-of-the-art performance in terms of prediction accuracy.

In this work, we systematically examine the application of spatio-temporal splitting heuristics to the Multi-Robot Motion Planning (MRMP) problem in a graph-theoretic setting: a problem known to be NP-hard to optimally solve. Following the divide-and-conquer principle, we design multiple spatial and temporal splitting schemes that can be applied to any existing MRMP algorithm, including integer programming solvers and Enhanced Conflict Based Search, in an orthogonal manner. The combination of a good baseline MRMP algorithm with a proper splitting heuristic proves highly effective, allowing the resolution of problems 10+ times than what is possible previously, as corroborated by extensive numerical evaluations. Notably, spatial partition of problem fusing with the temporal splitting heuristic and the enhanced conflict based search (ECBS) algorithm increases the scalability of ECBS on large and challenging DAO maps by 5–15 folds with negligible impact on solution optimality.

Robots that can manipulate objects in unstructured environments and collaborate with humans can benefit immensely by understanding natural language. We propose a pipelined architecture of two stages to perform spatial reasoning on the text input. All the objects in the scene are first localized, and then the instruction for the robot in natural language and the localized co-ordinates are mapped to the start and end co-ordinates corresponding to the locations where the robot must pick up and place the object respectively. We show that representing the localized objects by quantizing their positions to a binary grid is preferable to representing them as a list of 2D co-ordinates. We also show that attention improves generalization and can overcome biases in the dataset. The proposed method is used to pick-and-place playing cards using a robot arm.

The ability to communicate intention enables decentralized multi-agent robots to collaborate while performing physical tasks. In this work, we present spatial intention maps, a new intention representation for multi-agent vision-based deep reinforcement learning that improves coordination between decentralized mobile manipulators. In this representation, each agent’s intention is provided to other agents, and rendered into an overhead 2D map aligned with visual observations. This synergizes with the recently proposed spatial action maps framework, in which state and action representations are spatially aligned, providing inductive biases that encourage emergent cooperative behaviors requiring spatial coordination, such as passing objects to each other or avoiding collisions. Experiments across a variety of multi-agent environments, including heterogeneous robot teams with different abilities (lifting, pushing, or throwing), show that incorporating spatial intention maps improves performance for different mobile manipulation tasks while significantly enhancing cooperative behaviors.

Some impressing multi-kernel or multi-task correlation filter trackers only focus on boosting the discrimination of multi-channel features, or exploiting the interdependence among different tasks. However, the cooperation and complementary of both technologies are missed, and the spatial structure among or inside target regions is also ignored. Therefore, this paper proposes a spatial graph regularized hierarchical subtask multi-kernel cross-correlation tracker (GHMK) via Gaussian process regression view, which enjoys the merits of multi-subtask multi-kernel learning and Gaussian process regression to jointly learn kernel cross-correlation filters, and makes them complement and boost each other. The interdependence and discrimination among multi-kernel multi-subtask are jointly exploited by group structure sparsity, which is also used to evaluate spatial feature selection. The spatial graph is constructed via cross similarity to maintain the geometric structure among or inside hierarchy subtasks. Besides, the developed model is general, and provides a unified solution from GPR for CF trackers without boundary effect. Comprehensive experiments demonstrate its favorable and competitive performance against the state-of-the-art trackers.

Optimal control is a popular approach to synthesize highly dynamic motion. Commonly, L2 regularization is used on the control inputs in order to minimize energy used and to ensure smoothness of the control inputs. However, for some systems, such as satellites, the control needs to be applied in sparse bursts due to how the propulsion system operates. In this paper, we study approaches to induce sparsity in optimal control solutions—namely via smooth L1 and Huber regularization penalties. We apply these loss terms to state-of-the-art Differential Dynamic Programming (DDP)-based solvers to create a family of sparsity-inducing optimal control methods. We analyze and compare the effect of the different losses on inducing sparsity, their numerical conditioning, their impact on convergence, and discuss hyperparameter settings. We demonstrate our method in simulation and hardware experiments on canonical dynamics systems, control of satellites, and the NASA Valkyrie humanoid robot. We provide an implementation of our method and all examples for reproducibility on GitHub.

Sparse roadmaps are important to compactly represent state spaces, to determine problems to be infeasible and to terminate in finite time. However, sparse roadmaps do not scale well to high-dimensional planning problems. In prior work, we showed improved planning performance on high-dimensional planning problems by using multilevel abstractions to simplify state spaces. In this work, we generalize sparse roadmaps to multilevel abstractions by developing a novel algorithm, the sparse multilevel roadmap planner (SMLR). To this end, we represent multilevel abstractions using the language of fiber bundles, and generalize sparse roadmap planners by using the concept of restriction sampling with visibility regions. We argue SMLR to be probabilistically complete and asymptotically near-optimal by inheritance from sparse roadmap planners. In evaluations, we outperform sparse roadmap planners on challenging planning problems, in particular problems which are high-dimensional, contain narrow passages or are infeasible. We thereby demonstrate sparse multilevel roadmaps as an efficient tool for feasible and infeasible high-dimensional planning problems.

Autonomous systems often use approximate planners that exploit state abstractions to solve large MDPs in real-time decision-making problems. However, these planners can eliminate details needed to produce effective behavior in autonomous systems. We therefore propose a novel model, a partially abstract MDP, with a set of abstract states that each compress a set of ground states to condense irrelevant details and a set of ground states that expand from a set of expanded abstract states to retain relevant details. This papers offers (1) a definition of a partially abstract MDP that (2) generalizes its ground MDP and its abstract MDP and exhibits bounded optimality depending on its abstract MDP along with (3) a lazy algorithm for planning and execution in autonomous systems. The result is a scalable approach that computes near-optimal solutions to large problems in minutes rather than hours.

Visual place recognition is the task of finding matchings of images that show the same place in the world. Combinations of appearance changes (e.g. changing illumination or weather) and geometric changes (e.g. viewpoint changes or occlusions) challenge existing approaches. Learning-based local image feature pipelines are a promising approach to this type of problem. We present a novel attentive feature pooling method that can be used to train a CNN to jointly detect and describe local image features. It can be trained on small or moderately sized datasets with weak supervision in a classification training setup (e.g. we use a set of 24k images of publicly available web-camera images in our experiments). We propose to use a joint loss function that combines the cross-entropy loss for the classification task with a mean squared error in order to increase the repeatability of feature detections. We show how the approach can be integrated in a place recognition pipeline and run experiments on several standard place recognition datasets. Despite the small training dataset, we demonstrate a 15% improvement in the average performance compared to the best of a number of compared state-of-the-art approaches, and, probably more importantly, a 3x improvement in the worst-case performance. Open source code is available.

To design a general-purpose assembly robot system that can handle objects of various shapes, we propose a soft jig that fits to the shapes of assembly parts. The functionality of the soft jig is based on a jamming gripper developed in the field of soft robotics. The soft jig has a bag covered with a malleable silicone membrane, which has high friction, elongation, and contraction rates for keeping parts fixed. The bag is filled with glass beads to achieve a jamming transition. We propose a method to configure parts-fixing on the soft jig based on contact relations, reachable directions, and the center of gravity of the parts that are fixed on the jig. The usability of the soft jig was evaluated in terms of the fixing performance and versatility for various shapes and postures of parts.

Finite element methods have been successfully used to develop physics-based models of soft robots that capture the nonlinear dynamic behavior induced by continuous deformation. These high-fidelity models are therefore ideal for designing controllers for complex dynamic tasks such as trajectory optimization and trajectory tracking. However, finite element models are also typically very high-dimensional, which makes real-time control challenging. In this work we propose an approach for finite element model-based control of soft robots that leverages model order reduction techniques to significantly increase computational efficiency. In particular, a constrained optimal control problem is formulated based on a nonlinear reduced order finite element model and is solved via sequential convex programming. This approach is demonstrated through simulation of a cable-driven soft robot for a constrained trajectory tracking task, where a 9768-dimensional finite element model is used for controller design.

Unmanned aerial vehicles have been demonstrated successfully in a variety of tasks, including surveying and sampling tasks over large areas. These vehicles can take many forms. Quadrotors’ agility and ability to hover makes them well suited for navigating potentially tight spaces, while fixed wing aircraft are capable of efficient flight over long distances. Hybrid aerial vehicles (HAVs) attempt to achieve both of these benefits by exhibiting multiple modes; however, morphing HAVs typically require extra actuators which add mass, reducing both agility and efficiency. We propose a morphing HAV with folding wings that exhibits both a quadrotor and a fixed wing mode without requiring any extra actuation. This is achieved by leveraging the motion of a bistable mechanism at the center of the aircraft to drive folding of the wing using only the existing motors and the inertia of the system. We optimize both the bistable mechanism and the folding wing using a topology optimization approach. The resulting mechanisms were fabricated on a 3D printer and replaced the frame of an existing quadrotor. Our prototype successfully transitions between both modes and our experiments demonstrate that the behavior of the fabricated prototype is consistent with that of the simulation.

This paper considers the problem of multi-modal future trajectory forecast with ranking. Here, multi-modality and ranking refer to the multiple plausible path predictions and the confidence in those predictions, respectively. We propose Social-STAGE, Social interaction-aware Spatio-Temporal multi-Attention Graph convolution network with novel Evaluation for multi-modality. Our main contributions include analysis and formulation of multi-modality with ranking using interaction and multi-attention, and introduction of new metrics to evaluate the diversity and associated confidence of multi-modal predictions. We evaluate our approach on existing public datasets ETH and UCY and show that the proposed algorithm outperforms the state of the arts on these datasets.