ICRA 2023 论文列表

Springs can provide force at zero net energy cost by recycling negative mechanical work to benefit motor-driven robots or spring-augmented humans. However, humans have limited force and range of motion, and motors have a limited ability to produce force. These limits constrain how much energy a conventional spring can store and, consequently, how much assistance a spring can provide. In this paper, we introduce an approach to accumulating negative work in assistive springs over several motion cycles. We show that, by utilizing a novel floating spring mechanism, the weight of a human or robot can be used to iteratively increase spring compression, irrespective of the potential energy stored by the spring. Decoupling the force required to compress a spring from the energy stored by a spring advances prior works, and could enable spring-driven robots and humans to perform physically demanding tasks without the use of large actuators.

Achieving stable hopping has been a hallmark challenge in the field of dynamic legged locomotion. Controlled hopping is notably difficult due to extended periods of under-actuation combined with very short ground phases wherein ground interactions must be modulated to regulate global state. In this work, we explore the use of hybrid nonlinear model predictive control paired with a low-level feedback controller in a multi-rate hierarchy to achieve dynamically stable motions on a novel 3D hopping robot. In order to demonstrate richer behaviors on the manifold of rotations, both the planning and feedback layers must be designed in a geometrically consistent fashion; therefore, we develop the necessary tools to employ Lie group integrators and appropriate feedback controllers. We experimentally demonstrate stable 3D hopping on a novel robot, as well as trajectory tracking and flipping in simulation.

Noncontact particle manipulation (NPM) shows great application potential than its conventional counterpart particularly in terms of non-invasiveness, and thus has significantly extended robotic manipulation capacity into bio- medical engineering, material science, etc. As NPM by means of electric, magnetic, and optical field has successfully demonstrated powerful strength in both academia and industry, NPM boosted by acoustic field, however, still faces staggering challenges. It is indeed in the very recent years that controllable dynamic airborne or waterborne acoustic field modulation technology emerged in academia. In this paper, we report our latest research regarding dexterous and dynamic noncontact micro-particle manipulation on water surface effected by acoustic field in terms of automated trapping, closed-loop positioning, and real-time motion planning, which can be applied to scenarios such as parallel 3D printing, cell assembly, etc. The main contribution of this work is we demonstrated the feasibility of objective-oriented and fully automated acoustic manipulation of micro-particle in precision scale based on robotic approach in 2D plane. Experiment results showed that the repetitive positioning accuracy can reach as high as 16 μm, which is essentially the pixel scale factor.

This paper investigates the stochastic moving target encirclement problem in a realistic setting. In contrast to typical assumptions in related works, the target in our work is non-cooperative and capable of escaping the circle containment by boosting its speed to maximum for a short duration. In extreme conditions, where GPS signals are not available, weight restrictions are present, and ground guidance is absent, the agents can rely solely on their onboard single-modality perception tools to measure the distances to the target. The distance measurement allows for creating a position estimator by providing a target position-dependent variable. Furthermore, the construction of the unique distributed anti-synchronization controller (DASC) can guarantee that the two agents track and encircle the target swiftly. The convergence of the estimator and controller is rigorously evaluated using the Lyapunov technique. A real-world UAV-based experiment is conducted to illustrate the performance of the proposed methodology in addition to a simulated Matlab numerical sample. Our video demonstration can be found in the URL https://youtu.be/EDVLvP-bk8M.

Knowing the relative rotation angle improves relative pose estimation accuracy. We consider the problem of computing relative motion from a non-minimal number of correspondences with a known relative rotation angle. While several solvers for minimum correspondences have been proposed, no non-minimal solver for this problem currently exists. In this work, we propose two non-minimal solvers for this problem. The first solver solves the problem using convex relaxation and semidefinite programming, yielding certifiable solutions. The second method approaches the problem through local eigenvalue optimization with random initialization. Increasing the number of initial guesses lowers the chances of missing the correct solution. We conduct experiments on synthetic and real data, confirming our methods' advantages over competing methods.

Despite the growing interest in the use of drone fleets for delivery of food and parcels, the negative impact of such technology is still poorly understood. In this paper we investigate the impact of such fleets in terms of noise pollution and environmental justice. We use simulation with real population data to analyze the spatial distribution of noise, and find that: 1) noise increases rapidly with fleet size; and 2) drone fleets can produce noise hotspots that extend far beyond warehouses or charging stations, at levels that lead to annoyance and interference of human activities. This, we will show, leads to concerns of fairness of noise distribution. We then propose an algorithm that successfully balances the spatial distribution of noise across the city, and discuss the limitations of such purely technical approaches. We complement the work with a discussion of environmental justice, showing how careless UAV fleet development and regulation can lead to reinforcing well-being deficiencies of poor and marginalized communities.

It is well known that the singularities of a Gough-Stewart platform arise when the determinant of the Plücker coordinates of the robot leg lines vanish. The direct expansion of this determinant in terms of the configuration of the moving platform leads to an intimidating algebraic expression which is difficult to organize in a manner that facilitates extracting geometric conditions for singularities to occur. The use of Grassmann-Cayley algebra has permitted expressing this determinant as a bracket polynomial which is easier to manipulate symbolically. Each monomial in this polynomial is the product of three brackets, 4×4 determinants involving the homogeneous coordinates of four leg attachments. In this paper, we show how to derive, using elementary linear algebra arguments, bracket polynomials where all brackets can be interpreted as reciprocal products between lines. Contrarily to what one might expect, these new bracket polynomials are simpler in general than those previously obtained using Grassmann-Cayley algebra.

With the advancement of industrial manufacturing and an increase in introduction of robots in the workspace, the need of safe operation, communication and information sharing is paramount. The work presented here focuses on cyber-physical system integration through Digital Twin (DT) technology. Our novel DT architecture is based on a model-free Neuro-Adaptive controller (NAC), and an edge-computing scheme for scene monitoring. The NAC can account for varying robot dynamics in both real and virtual environments, and allows for the DT system to expand the realm of cyber-physical integration without expensive model tuning. The edge-computing device introduced in our architecture, observes the robot's workspace from a distance with a wider field of view. This wide viewpoint, enhances the detection and mitigation of any obstacles entering the robot's workspace during operation. We experimentally evaluated the performance of our proposed architecture by introducing dynamic obstacles during a pick-and-place task that both the physical robot and its digital twin had to avoid. Results show that the proposed DT architecture successfully integrates the novel controller and edge-computing elements and successfully performs the given navigation task. The results also show that NAC outperforms a PD controller with more than 70% improvement in joint tracking error between the physical and virtual robots. It was observed that the latency experienced while using NAC is about 48 % lower than when Proportional-Derivative (PD) controller was operational.

Precision agricultural robots require high-resolution navigation solutions. In this paper, we introduce a robust neural-inertial sequence learning approach to track such robots with ultra-intermittent GNSS updates. First, we propose an ultra-lightweight neural-Kalman filter that can track agricultural robots within 1.4 m (1.4–5.8× better than competing techniques), while tracking within 2.75 m with 20 mins of GPS outage. Second, we introduce a user-friendly video-processing toolbox to generate high-resolution (±5 cm) position data for fine-tuning pre-trained neural-inertial models in the field. Third, we introduce the first and largest (6.5 hours, 4.5 km, 3 phases) public neural-inertial navigation dataset for precision agricultural robots. The dataset, toolbox, and code are available at: https://github.com/nesl/agrobot.

We study the problem of generating control laws for systems with unknown dynamics. Our approach is to represent the controller and the value function with neural networks, and to train them using loss functions adapted from the Hamilton-Jacobi-Bellman (HJB) equations. In the absence of a known dynamics model, our method first learns the state transitions from data collected by interacting with the system in an offline process. The learned transition function is then integrated to the HJB equations and used to forward simulate the control signals produced by our controller in a feedback loop. In contrast to trajectory optimization methods that optimize the controller for a single initial state, our controller can generate near-optimal control signals for initial states from a large portion of the state space. Compared to recent model-based reinforcement learning algorithms, we show that our method is more sample efficient and trains faster by an order of magnitude. We demonstrate our method in a number of tasks, including the control of a quadrotor with 12 state variables.

We present a technique for dense 3D reconstruction of objects using an imaging sonar, also known as forward-looking sonar (FLS). Compared to previous methods that model the scene geometry as point clouds or volumetric grids, we represent the geometry as a neural implicit function. Additionally, given such a representation, we use a differentiable volumetric renderer that models the propagation of acoustic waves to synthesize imaging sonar measurements. We perform experiments on real and synthetic datasets and show that our algorithm reconstructs high-fidelity surface geometry from multi-view FLS images at much higher quality than was possible with previous techniques and without suffering from their associated memory overhead.

We formulate grasp learning as a neural field and present Neural Grasp Distance Fields (NGDF). Here, the input is a 6D pose of a robot end effector and output is a distance to a continuous manifold of valid grasps for an object. In contrast to current approaches that predict a set of discrete candidate grasps, the distance-based NGDF representation is easily interpreted as a cost, and minimizing this cost produces a successful grasp pose. This grasp distance cost can be incorporated directly into a trajectory optimizer for joint optimization with other costs such as trajectory smoothness and collision avoidance. During optimization, as the various costs are balanced and minimized, the grasp target is allowed to smoothly vary, as the learned grasp field is continuous. We evaluate NGDF on joint grasp and motion planning in simulation and the real world, outperforming baselines by 63 % execution success while generalizing to unseen query poses and unseen object shapes. Project page: https://sites.google.com/view/neural-grasp-distance-fields.

We present Neural Contact Fields, a method that brings together neural fields and tactile sensing to address the problem of tracking extrinsic contact between object and environment. Knowing where the external contact occurs is a first step towards methods that can actively control it in facilitating downstream manipulation tasks. Prior work for localizing environmental contacts typically assume a contact type (e.g. point or line), does not capture contact/no-contact transitions, and only works with basic geometric-shaped objects. Neural Contact Fields are the first method that can track arbitrary multi-modal extrinsic contacts without making any assumptions about the contact type. Our key insight is to estimate the probability of contact for any 3D point in the latent space of object's shapes, given vision-based tactile inputs that sense the local motion resulting from the external contact. In experiments, we find that Neural Contact Fields are able to localize multiple contact patches without making any assumptions about the geometry of the contact, and capture contact/no-contact transitions for known categories of objects with unseen shapes in unseen environment configurations. In addition to Neural Contact Fields, we also release our YCB-Extrinsic-Contact dataset of simulated extrinsic contact interactions to enable further research in this area. Project page: https://github.com/carolinahiguera/NCF

Most recently proposed methods for robotic per-ception are based on deep learning, which require very large datasets to perform well. The accuracy of a learned model is mainly dependent on the data distribution it was trained on. Thus for deploying such models, it is crucial to use training data belonging to the robot's environment. However, collecting and labeling data is a significant bottleneck, necessitating efficient data collection and labeling pipelines. This paper presents a method to compute high-quality object segmentation maps for RGB-D video sequences using minimal human labeling effort. We leverage the density learned by a Neural Radiance Field (NeRF) to infer the geometry of the scene, which we use to compute dense segmentation maps using a single 3D bounding box provided by a user. We study the accuracy of the computed segmentation maps and present a way to generate additional synthetic training examples observing the scene from novel viewpoints using the learned radiance fields. Our results show that our method is able to compute accurate segmentation maps, outperforming baseline and state-of-the-art methods. We also show that using the synthetic training examples improves performance on a downstream object detection task.

We present a system for applying sim2real approaches to “in the wild” scenes with realistic visuals, and to policies which rely on active perception using RGB cameras. Given a short video of a static scene collected using a generic phone, we learn the scene's contact geometry and a function for novel view synthesis using a Neural Radiance Field (NeRF). We augment the NeRF rendering of the static scene by overlaying the rendering of other dynamic objects (e.g. the robot's own body, a ball). A simulation is then created using the rendering engine in a physics simulator which computes contact dynamics from the static scene geometry (estimated from the NeRF vol-ume density) and the dynamic objects' geometry and physical properties (assumed known). We demonstrate that we can use this simulation to learn vision-based whole body navigation and ball pushing policies for a 20 degree-of-freedom humanoid robot with an actuated head-mounted RGB camera, and we successfully transfer these policies to a real robot.

We propose a novel visual re-localization method based on direct matching between the implicit 3D descriptors and the 2D image with transformer. A conditional neural radiance field(NeRF) is chosen as the 3D scene representation in our pipeline, which supports continuous 3D descriptors generation and neural rendering. By unifying the feature matching and the scene coordinate regression to the same framework, our model learns both generalizable knowledge and scene prior respectively during two training stages. Furthermore, to improve the localization robustness when domain gap exists between training and testing phases, we propose an appearance adaptation layer to explicitly align styles between the 3D model and the query image. Experiments show that our method achieves higher localization accuracy than other learning-based approaches on multiple benchmarks. Code is available at https://github.com/JenningsL/nerf-loc.

This paper firstly presents our optimal path planning algorithm within a 2\mathrm{D}2\mathrm{D} non-convex, polytopic region defined as a sequence of connected convex polytopes. The path is a B-spline curve but being parametrized with its equivalent Bézier representation. By doing this, the local convexity bound of each curve's interval is significantly tighter. Thus, it allows many more possibilities for constraining the entire curve to remain inside the region by using only linear constraints on the control points of the curve. We further guarantee the existence of the valid path by pointing out an algebraic solution. We integrate the algorithm, together with our previously published results, into the Navigation with polytopes toolbox which can be used as a global path planner, compatible with ROS navigation tools. It provides a framework for constructing a polytope map from a standard occupancy gridmap, searching for an appropriate sequence of connected polytopes and finally, planning a minimal-length path with different options on B-spline or Bézier parametrizations. The validation and comparison with existing methods are done using gridmaps collected under Gazebo simulations and real experiments.

Recent work on battery-free soft robotics has demonstrated the use of liquid crystal elastomers (LCE) to build shape-changing materials activated by applied external heat. However, sources of heat must typically be in direct field-of-view of the robot (i.e. NIR, laser, and visual light EM sources or convective heats guns), be tethered to an external power supply (i.e. thermoelectric heating or resistive joule heaters), or require a heavy on-board battery that limits mobility and range. This paper presents a novel battery-free soft-robotics platform that can crawl through confined, enclosed, and hard-to-reach spaces (e.g. packages, machinery, pipes, etc.), hidden from view of heating infrastructure. This is achieved through the co-design of a soft robotics platform and integrated soft conductive traces that enable wireless (microwave) heating through remote stimulation. We achieve fast actuation through a careful choice of materials and the overall mechanical structure of the robot to maximize heating efficiency. Further, the robot is actively tracked through enclosed spaces using a mm Wave radar to direct heat to its location. We provide a detailed evaluation on the robot's heating efficiency, location-tracking accuracy and crawling speed.

It has always been expected that the robot can understand the natural language instruction and thus a more natural human-robot interaction is achieved. Currently, the robot usually interprets the instruction by visually grounding the textual information to its surroundings, while it may be not enough for some complex situations with only visual perception. So it is reasonable for the robot to leverage its multisensory perception ability to better understand the instruction. In this paper, we propose a multisensory perception approach to tackle the task of natural language instruction understanding for robotic manipulation, in which the robot coordinates its visual, tactile and auditory perception to fully understand the instruction and then executes the manipulation task. Extensive experiments have been conducted demonstrating the superiority of the multisensory perception compared with single sensory perception for instruction understanding. Moreover, we establish a user-friendly human-robot interaction interface where the human sends instruction to the robot via a mobile APP.