IROS 2023 论文列表

In the last decade, researchers have been trying to develop many microrobots that mimic the extraordinary abilities of bionts in complex environments. How to fabricate the biomimetic microrobot with satisfying deformability and complex shapes to realize desired precise motion is the key issue. In this paper, we proposed an efficient programable fabrication method of the magnetic soft micro-robot through an on-chip photopolymerization system. The superparamagnetic nanoparticles were compiled according to the magnetic anisotropy and assembled in the micro-robot. Then these nanoparticles were immobilized by photopolymerization of the hydrogel polymer. With this fabrication method, a joint rotation mechanism was first fabricated to characterize the deformation performance under the magnetic field control. Besides, the snake-like micro-robot were also fabricated, and the desired motions were achieved. The experimental results show that the proposed programable on-chip fabrication of magnetic soft micro-robot has the potential to facilitate the development of magnetic microrobots and their applications in the biomedical field.

A key challenge in off-road navigation is that even visually similar terrains or ones from the same semantic class may have substantially different traction properties. Existing work typically assumes no wheel slip or uses the expected traction for motion planning, where the predicted trajectories provide a poor indication of the actual performance if the terrain traction has high uncertainty. In contrast, this work proposes to analyze terrain traversability with the empirical distribution of traction parameters in unicycle dynamics, which can be learned by a neural network in a self-supervised fashion. The probabilistic traction model leads to two risk-aware cost formulations that account for the worst-case expected cost and traction. To help the learned model generalize to unseen environment, terrains with features that lead to unreliable predictions are detected via a density estimator fit to the trained network's latent space and avoided via auxiliary penalties during planning. Simulation results demonstrate that the proposed approach outperforms existing work that assumes no slip or uses the expected traction in both navigation success rate and completion time. Furthermore, avoiding terrains with low density-based confidence score achieves up to 30% improvement in success rate when the learned traction model is used in a novel environment.

To safely and efficiently extract an object from the clutter, this paper presents a bimanual manipulation planner in which one hand of the robot is used to slide the target object out of the clutter while the other hand is used to support the surrounding objects to prevent the clutter from collapsing. Our method uses a neural network to predict the physical phenomena of the clutter when the target object is moved. We generate the most efficient action based on the Monte Carlo tree search. The grasping and sliding actions are planned to minimize the number of motion sequences to pick the target object. In addition, the object to be supported is determined to minimize the position change of surrounding objects. Experiments with a real bimanual robot confirmed that the robot could retrieve the target object, reducing the total number of motion sequences and improving safety.

Leveraging recent developments in black-box risk-aware verification, we provide three algorithms that generate probabilistic guarantees on (1) optimality of solutions, (2) recursive feasibility, and (3) maximum controller runtimes for general nonlinear safety-critical finite-time optimal controllers. These methods forego the usual (perhaps) restrictive assumptions required for typical theoretical guarantees, e.g. terminal set calculation for recursive feasibility in Nonlinear Model Predictive Control, or convexification of optimal controllers to ensure optimality. Furthermore, we show that these methods can directly be applied to hardware systems to generate controller guarantees on their respective systems.

We propose, analyze, and experimentally verify a new proactive approach for robot social navigation driven by the robot's “opinion” for which way and by how much to pass human movers crossing its path. The robot forms an opinion over time according to nonlinear dynamics that depend on the robot's observations of human movers and its level of attention to these social cues. For these dynamics, it is guaranteed that when the robot's attention is greater than a critical value, deadlock in decision making is broken, and the robot rapidly forms a strong opinion, passing each human mover even if the robot has no bias nor evidence for which way to pass. We enable proactive rapid and reliable social navigation by having the robot grow its attention across the critical value when a human mover approaches. With human-robot experiments we demonstrate the flexibility of our approach and validate our analytical results on deadlock-breaking. We also show that a single design parameter can tune the trade-off between efficiency and reliability in human-robot passing. The new approach has the additional advantage that it does not rely on a predictive model of human behavior.

For robots navigating in dynamic environments, exploiting and understanding uncertain human motion prediction is key to generate efficient, safe and legible actions. The robot may perform poorly and cause hindrances if it does not reason over possible, multi-modal future social interactions. With the goal of enhancing autonomous navigation in cluttered environments, we propose a novel formulation for nonlinear model predictive control including multi-modal predictions of human motion. As a result, our approach leads to less conservative, smooth and intuitive human-aware navigation with reduced risk of collisions, and shows a good balance between task efficiency, collision avoidance and human comfort. To show its effectiveness, we compare our approach against the state of the art in crowded simulated environments, and with real-world human motion data from the THOR dataset. This comparison shows that we are able to improve task efficiency, keep a larger distance to humans and significantly reduce the collision time, when navigating in cluttered dynamic environ-ments. Furthermore, the method is shown to work robustly with different state-of-the-art human motion predictors.

Deep learning is the method of choice for trajectory prediction for autonomous vehicles. Unfortunately, its data-hungry nature implicitly requires the availability of sufficiently rich and high-quality centralized datasets, which easily leads to privacy leakage. Besides, uncertainty-awareness becomes increasingly important for safety-crucial cyber physical systems whose prediction module heavily relies on machine learning tools. In this paper, we relax the data collection requirement and enhance uncertainty-awareness by using Federated Learning on Connected Autonomous Vehicles with an uncertainty-aware global objective. We name our algorithm as FLTP. We further introduce ALFLTP which boosts FLTP via using active learning techniques in adaptatively selecting participating clients. We consider two different metrics negative log-likelihood (NLL) and aleatoric uncertainty (AU) for client selection. Experiments on Argoverse dataset show that FLTP significantly outperforms the model trained on local data. In addition, ALFLTP-AU converges faster in training regression loss and performs better in terms of Miss Rate (MR) than FLTP in most rounds, and has more stable round-wise performance than ALFLTP-NLL.

In scenarios involving grasping multiple targets, the learning of stacking relationships between objects is fundamental for robots to execute safely and efficiently. However, current methods lack subdivision for the hierarchy of stacking relationship types. In scenes where objects are mostly stacked in an orderly manner, they are incapable of performing human-like and high-efficient grasping decisions. This paper proposes a perception-planning method to distinguish different stacking forms between objects and generate prioritized manipulation sequences based on given target designations. We utilize a Hierarchical Stacking Relationship Network (HSRN) to discriminate the hierarchy of stacking and generate a refined Stacking Relationship Tree (SRT) for relationship description. Considering objects with high stacking stability can be processed together if necessary, we introduce an elaborate decision-making planner based on Partially Observable Markov Decision Process (POMDP), which leverages observations and generates the least grasp-consuming decision chain with robustness and is suitable for simultaneously specifying multiple targets. To verify our work, we set the scene to the dining table and augment REGRAD dataset for network training. Experiments show that our method effectively generates grasping decisions that conform to human requirements, and improves the implementation efficiency compared with existing methods on the basis of guaranteeing success rate.

Soft robots made of flexible materials are highly adaptive, easy to fabricate, and safer to interact with. One of the ways for soft robots to interact with the surrounding environment is through their deformable bodily characteristics including internal body stiffness and external body friction. Though the flexibility of soft-bodied robots has been rigorously studied, the frictional skin of such soft-bodied robots, acting as a mechanical interface between the robot and the environment, remains unexplored. Being able to design the frictional skin will make soft-bodied robots more versatile in environmental navigation, more dexterous in manipulation tasks, and more flexible in haptic feedback. In this paper, we propose a robotic skin that can be programmed dynamically to change the mode of friction. The robotic skin is based on bistable bellow structures that can be switched between two folding states to change the contact points between the robotic skin and the ground. Our robotic skin can dynamically change its anisotropic frictional behavior to add another dimension to the designing space of soft robotics.

The Exobiology Extant Life Surveyor (EELS) is a snake-like mobile instruments platform under development at Jet Propulsion Laboratory (JPL) for a mission concept to find evidence of life on Saturn's sixth largest moon, Enceladus. To conduct a life surveying mission there, the EELS platform must first traverse an unknown icy surface terrain before undertaking a controlled descent into a cryovolcanic vent. The remoteness of Enceladus and the icy nature of its terrain demands a level of autonomy in navigation significantly higher than previous rover missions. The perception system onboard EELS must be highly resilient to perceptually-degraded environments such as flat, open ice fields, icy plumes, and repeating geometries in vents. EELS' perception system is implemented as a multi-sensor Simultaneous Localisation And Mapping (SLAM) solution called SERPENT. State Estimation through Robust Perception in Extreme and Novel Terrains (SERPENT) estimates the robot trajectory and maintains a map database, from which dense global or local maps can be obtained on demand for downstream planning algorithms. This system opts to incorporate measurements from many sensor modalities (laser scans, images, IMU, altimeter, etc.), solving the SLAM problem through joint optimisation, and thus requires that the contribution of each sensor be balanced through careful modelling of their uncertainties. With a specific focus on Light Detection And Ranging (LiDAR) in this context, this paper proposes a principled approach to model the covariances of point-to-plane Iterative Closest Point (ICP). It performs a rigorous comparative analysis of new and existing covariance models, and is the first time some of these have been tested within a complete SLAM pipeline. These models are evaluated on perceptually challenging datasets collected in glacial environments by the EELS sensor suite (see Figures 1, 2). SERPENT is open-sourced at https://github.com/jpl-eels/serpent. Show More

Reinforcement learning (RL) algorithms face significant challenges when dealing with long-horizon robot manipulation tasks in real-world environments due to sample inefficiency and safety issues. To overcome these challenges, we propose a novel framework, SEED, which leverages two approaches: reinforcement learning from human feedback (RLHF) and primitive skill-based reinforcement learning. Both approaches are particularly effective in addressing sparse reward issues and the complexities involved in long-horizon tasks. By combining them, SEED reduces the human effort required in RLHF and increases safety in training robot manipulation with RL in real-world settings. Additionally, parameterized skills provide a clear view of the agent's high-level intentions, allowing humans to evaluate skill choices before they are executed. This feature makes the training process even safer and more efficient. To evaluate the performance of SEED, we conducted extensive experiments on five manipulation tasks with varying levels of complexity. Our results show that SEED significantly outperforms state-of-the-art RL algorithms in sample efficiency and safety. In addition, SEED also exhibits a substantial reduction of human effort compared to other RLHF methods. Further details and video results can be found at https://seediros23.github.io/.

The outdoor navigation capabilities of ground robots have improved significantly in recent years, opening up new potential applications in a variety of settings. Cost-based representations of the environment are frequently used in the path planning domain to obtain an optimized path based on various objectives, such as traversal time or energy consumption. However, obtaining such cost representations is still cumbersome, particularly in outdoor settings with diverse terrain types and slope angles. In this paper, we address this problem by using a data-driven approach to develop a cost representation for various outdoor terrain types that supports two optimization objectives, namely energy consumption and traversal time. We train a supervised machine learning model whose inputs consists of extracted environment data along a path and whose outputs are the predicted energy consumption and traversal time. The model is based on a ResNet neural network architecture and trained using field-recorded data. The error of the proposed method on different types of terrain is within 11% of the ground truth data. To show that it performs and generalizes better than currently existing approaches on various types of terrain, a comparison to a baseline method is made.

Robotic manipulation tasks rely on a plethora of environmental and payload information. One critical piece of information for accurate manipulation is the center of mass (CoM) of the object, which is essential for estimating the dynamic response of the system and determining the payload placement. Traditionally, the CoM of a payload is provided prior to manipulation. In order to create a more robust and comprehensive system, this information should be collected by the robotic agent before or during the task run time. This paper presents a method for approximating the CoM of a planar object using a small-scale mobile robot to inform manipulation tasks. On average, our system is able to converge on a CoM estimate in under 30 seconds in simulation and 20 seconds in experiment, with a relative error of 4.95% and 5.46%, respectively.

Prediction-based active perception has shown the potential to improve the navigation efficiency and safety of the robot by anticipating the uncertainty in the unknown environment. The existing works for 3D shape prediction make an implicit assumption about the partial observations and therefore cannot be used for real-world planning and do not consider the control effort for next-best-view planning. We present Pred-NBV, a realistic object shape reconstruction method consisting of PoinTr-C, an enhanced 3D prediction model trained on the ShapeNet dataset, and an information and control effort-based next-best-view method to address these issues. Pred-NBV shows an improvement of 25.46% in object coverage over the traditional methods in the AirSim simulator, and performs better shape completion than PoinTr, the state-of-the-art shape completion model, even on real data obtained from a Velodyne 3D LiDAR mounted on DJI M600 Pro.

We present a system for non-prehensile manipulation that require a significant number of contact mode transitions and the use of environmental contacts to successfully manipulate an object to a target location. Our method is based on deep reinforcement learning which, unlike state-of-the-art planning algorithms, does not require apriori knowledge of the physical parameters of the object or environment such as friction coefficients or centers of mass. The planning time is reduced to the simple feed-forward prediction time on a neural network. We propose a computational structure, action space design, and curriculum learning scheme that facilitates efficient exploration and sim-to-real transfer. In challenging real-world non-prehensile manipulation tasks, we show that our method can generalize over different objects, and succeed even for novel objects not seen during training. Project website: https://sites.google.com/view/nonprenehsile-decomposition

Ambulation in everyday life requires walking at variable speeds, variable inclines, and variable terrains. Powered prostheses aim to provide this adaptability through control of the actuated joints. Some powered prosthesis controllers can adapt to discrete changes in speed and incline but require manual tuning to determine the control parameters, leading to poor clinical viability. Other data-driven controllers can continuously adapt to changes in speed and incline but do so by imposing the same non-amputee gait patterns for all amputee subjects, which does not consider subjective preferences and differing clinical needs of users. Here, we present a controller for powered knee and ankle prostheses that can continuously adapt to different walking speeds, inclines, and uneven terrains without enforcing a specific prosthesis position, impedance, or torque. A virtual biarticular muscle connection determines the knee flexion torque, which changes with both speed and slope. Adaptation to inclines and uneven terrains is based solely on the global shank orientation. Continuously variable damping allows for speed adaptation. Minimum-jerk programming defines the prosthesis swing trajectory at variable cadences. Experiments with one individual with an above-knee amputation suggest that the proposed controller can effectively adapt to different walking speeds, inclines, and rough terrains.

Accurate estimation of the relative pose between an object and a robot hand is critical for many manipulation tasks. However, most of the existing object-in-hand pose datasets use two-finger grippers and also assume that the object remains fixed in the hand without any relative movements, which is not representative of real-world scenarios. To address this issue, a 6D object-in-hand pose dataset is proposed using a teleoperation method with an anthropomorphic Shadow Dexterous hand. Our dataset comprises RGB-D images, proprioception and tactile data, covering diverse grasping poses, finger contact states, and object occlusions. To overcome the significant hand occlusion and limited tactile sensor contact in real-world scenarios, we propose PoseFusion, a hybrid multi-modal fusion approach that integrates the information from visual and tactile perception channels. PoseFusion generates three candidate object poses from three estimators (tactile only, visual only, and visuo-tactile fusion), which are then filtered by a SelectLSTM network to select the optimal pose, avoiding inferior fusion poses resulting from modality collapse. Extensive experiments demonstrate the robustness and advantages of our framework. All data and codes are available on the project website: https://elevenjiang1.github.io/ObjectlnHand-Dataset/.

In recent years, there is a noteworthy advance-ment in autonomous drone racing. However, the primary focus is on attaining execution times, while scant attention is given to the challenges of dynamic environments. The high-speed nature of racing scenarios, coupled with the potential for unforeseeable environmental alterations, present stringent requirements for online replanning and its timeliness. For racing in dynamic environments, we propose an online replanning framework with an efficient polynomial trajectory representation. We trade off between aggressive speed and flexible obstacle avoidance based on an optimization approach. Additionally, to ensure safety and precision when crossing intermediate racing waypoints, we formulate the demand as hard constraints during planning. For dynamic obstacles, parallel multi-topology trajectory planning is designed based on engineering considerations to prevent racing time loss due to local optimums. The framework is integrated into a quadrotor system and successfully demonstrated at the DJI Robomaster Intelligent UAV Championship, where it successfully complete the racing track and placed first, finishing in less than half the time of the second-place11https://pro-robomasters-hz-n5i3.oss-cn-hangzhou.aliyuncs.com/sass/event-list.html.

Human skin can accurately sense the self-decoupled normal and shear forces when in contact with objects of different sizes. Although there exist many soft and conformable tactile sensors on robotic applications able to decouple the normal force and shear forces, the impact of the size of object in contact on the force calibration model has been commonly ignored. Here, using the principle that contact force can be derived from the light power loss in the soft optical fiber core, we present a soft tactile sensor that decouples normal and shear forces and calibrates the measurement results based on the object size, by designing a two-layered weaved polymer-based optical fiber anisotropic structure embedded in a soft elastomer. Based on the anisotropic response of optical fibers, we developed a linear calibration algorithm to simultaneously measure the size of the contact object and the decoupled normal and shear forces calibrated the object size. By calibrating the sensor at the robotic arm tip, we show that robots can reconstruct the force vector at an average accuracy of 0.15N for normal forces, 0.17N for shear forces in X-axis, and 0.18N for shear forces in Y-axis, within the sensing range of 0-2N in all directions, and the average accuracy of object size measurement of 0.4mm, within the test indenter diameter range of 5-12mm.