ICRA 2022 Papers

Multi-agent formation as well as obstacle avoid-ance is one of the most actively studied topics in the field of multi-agent systems. Although some classic controllers like model predictive control (MPC) and fuzzy control achieve a certain measure of success, most of them require precise global information which is not accessible in harsh environments. On the other hand, some reinforcement learning (RL) based approaches adopt the leader-follower structure to organize different agents' behaviors, which sacrifices the collaboration between agents thus suffering from bottlenecks in maneuver-ability and robustness. In this paper, we propose a distributed formation and obstacle avoidance method based on multi-agent reinforcement learning (MARL). Agents in our system only utilize local and relative information to make decisions and control themselves distributively, and will reorganize themselves into a new topology quickly in case that any of them is dis-connected. Our method achieves better performance regarding formation error, formation convergence rate and on-par success rate of obstacle avoidance compared with baselines (both classic control methods and another RL-based method). The feasibility of our method is verified by both simulation and hardware implementation with Ackermann-steering vehicles.

Picking cluttered general objects is a challenging task due to the complex geometries and various stacking configurations. Many prior works utilize pose estimation for picking, but pose estimation is difficult on cluttered objects. In this paper, we propose Cluttered Objects Descriptors (CODs), a dense cluttered objects descriptor which can represent rich object structures, and use the pre-trained CODs network along with its intermediate outputs to train a picking policy. Additionally, we train the policy with reinforcement learning, which enable the policy to learn picking without supervision. We conduct experiments to demonstrate that our CODs is able to consistently represent seen and unseen cluttered objects, which allowed for the picking policy to robustly pick cluttered general objects. The resulting policy can pick 96.69% of unseen objects in our experimental environment that are twice as cluttered as the training scenarios.

CPG-Based oscillator networks are increasingly being used to drive multi-limbed robots. To produce effective gaits with these networks, the relationship between the CPG parameters and the characteristics of the gait must be determined. However, due to the nonlinear nature of the oscillators, this relationship is challenging to ascertain. In this work a reinforcement learning algorithm is used to determine the CPG parameters that produce propulsively beneficial kinematics in a multi-fin underwater robot. Due to the high computational cost in creating high fidelity simulations of underwater systems, an alternate method using a low fidelity simulation is explored. To better simulate the dynamics of a two-finned swimming robot a thorough force sweep is conducted on the subject robot in a controlled environment. The resulting force data is used as the dynamic information in a simple simulation. This method allows for the learning of CPG weight settings that produce desired kinematic operating conditions and their resulting forces in simulation. Using this method, when the learned CPG parameters were applied directly to the physical robot, the robot executed the same desired kinematics and forces as expected from simulation with no additional learning needed.

In unstructured urban canals, regulation-aware interactions with other vessels are essential for collision avoidance and social compliance. In this paper, we propose a regulations aware motion planning framework for Autonomous Surface Vessels (ASVs) that accounts for dynamic and static obstacles. Our method builds upon local model predictive contouring control (LMPCC) to generate motion plans satisfying kino-dynamic and collision constraints in real-time while including regulation awareness. To incorporate regulations in the planning stage, we propose a cost function encouraging compliance with rules describing interactions with other vessels similar to COLlision avoidance REGulations at sea (COLREGs). These regulations are essential to make an ASV behave in a predictable and socially compliant manner with regard to other vessels. We compare the framework against baseline methods and show more effective regulation-compliant avoidance of moving obstacles with our motion planner. Additionally, we present experimental results in an outdoor environment.

The image signal processing (ISP) pipeline, which transforms raw sensor measurement to a color image, is composed of a sequence of processing modules. Traditionally, the ISP pipeline is manually tuned by experts for human perception. The resulting handcrafted ISP configuration does not necessarily benefit the downstream high-level vision tasks. To mitigate these problems, this paper presents a simple yet effective framework based on Evolutionary Algorithm to search for a set of compact ISP configurations for high-level vision tasks. In particular, we encode ISP structure into a binary string and ISP parameters into a set of float numbers. Then we jointly optimize them with task-specific loss and ISP computation budgets (e.g., running time) through solving a nonlinear multi-objective optimization problem. By mutating the configurations of the ISP pipeline, we are able to remove redundant modules and design an ISP with both low cost and high accuracy. We validate the proposed method on extreme noisy and low-light raw images, and experimental results show that our framework can help find effective and efficient ISP configurations for both object detection and semantic segmentation tasks. We further provide a detailed analysis on the importance of different modules in the ISP configurations, which benefits the design of ISP for downstream tasks in the future.

Recently simulation methods have been developed for optical tactile sensors to enable the Sim2Real learning, i.e., first training models in simulation before deploying them on a real robot. However, some artefacts in real objects are unpredictable, such as imperfections caused by fabrication processes, or scratches by natural wear and tear, and thus cannot be represented in the simulation, resulting in a significant gap between the simulated and real tactile images. To address this Sim2Real gap, we propose a novel texture generation network to map the simulated images into photorealistic tactile images that resemble a real sensor contacting a real imperfect object. Each simulated tactile image is first divided into two types of regions: areas that are in contact with the object and areas that are not. The former is applied with generated textures learned from real textures in the real tactile images, whereas the latter maintains its appearance as when the sensor is not in contact with any object. This makes sure that the artefacts are only applied to deformed regions of the sensor. Our extensive experiments show that the proposed texture generation network can generate realistic artefacts on the deformed regions of the sensor, while avoiding leaking the textures into areas of no contact. Quantitative experiments further reveal that when using the adapted images generated by our proposed network for a Sim2Real classification task, the drop in accuracy caused by the Sim2Real gap is reduced from 38.43% to merely 0.81%. As such, this work has potential to accelerate the Sim2Real learning for robotic tasks requiring tactile sensing.

Quadratic Program(QP) based state-feedback controllers, whose inequality constraints bound the rate of change of control barrier (CBFs) and lyapunov function with a class-\mathcal{K}\mathcal{K} function of their values, are sensitive to the parameters of these class-\mathcal{K}\mathcal{K} functions. The construction of valid CBFs, however, is not straightforward, and for arbitrarily chosen parameters of the QP, the system trajectories may enter states at which the QP either eventually becomes infeasible, or may not achieve desired performance. In this work, we pose the control synthesis problem as a differential policy whose parameters are optimized for performance over a time horizon at high level, thus resulting in a bi-level optimization routine. In the absence of knowledge of the set of feasible parameters, we develop a Recursive Feasibility Guided Gradient Descent approach for updating the parameters of QP so that the new solution performs at least as well as previous solution. By considering the dynamical system as a directed graph over time, this work presents a novel way of optimizing performance of a QP controller over a time horizon for multiple CBFs by (1) using the gradient of its solution with respect to its parameters by employing sensitivity analysis, and (2) backpropagating these as well as system dynamics gradients to update parameters while maintaining feasibility of QPs.

Anthropomorphic robotic grippers are required for robots, prostheses, and orthosis to enable manipulation of a priori unknown and variable-shape objects. It has to meet a wide range of sometimes contradictory requirements in terms of adaptivity, dexterity, high payload to weight ratio, robustness, aesthetics, compactness, lightweight, etc. Within this paper, we utilize the morphological computation approach to introduce design for anthropomorphic re-configurable underactuated grippers. The key to fingers' adaptivity is embedded passive variable length links and elastic elements at input joints. Based on this concept, we designed a palm-size five-finger gripper, where 14 DoFs, including thumb, are controlled by just 4 motors, such that it can perform both precision pinch and encompassing power grasps of various objects. The paper describes synthesized linkages for digits, hand design overview, control strategy, and test results of a physical prototype.

Planning the motion of bearings-only sensors is critical for enabling accurate tracking of the positions of moving targets. In this paper, we demonstrate planning the observer's motion over horizons greater than one step for estimating an unknown and varying number of indistinguishable, maneuvering targets of interest using a probability hypothesis density (PHD) filter, with a Rériyi divergence reward for selecting actions. We describe approximations to make this approach computationally feasible, and we propose using Monte Carlo tree search (MCTS) to further reduce the cost. Finally, we present simulation results showing that longer planning horizons reduce the error in the estimates and that MCTS can reduce the cost of planning without sacrificing the quality of the estimates.

In order to safely and efficiently collaborate with humans, industrial robots need the ability to alter their motions quickly to react to sudden changes in the environment, such as an obstacle appearing across a planned trajectory. In Realtime Motion Planning, obstacles are detected in real time through a vision system, and new trajectories are planned with respect to the current positions of the obstacles, and immediately executed on the robot. Existing realtime motion planners, however, lack the smoothing post-processing step - which are crucial in sampling-based motion planning - resulting in the planned trajectories being jerky, and therefore inefficient and less human-friendly. Here we propose a Realtime Trajectory Smoother based on the shortcutting technique to address this issue. Leveraging fast clearance inference by a novel neural network, the proposed method is able to consistently smooth the trajectories of a 6-DOF industrial robot arm within 200 ms on a commercial GPU. We integrate the proposed smoother into a full Vision-Motion Planning-Execution loop and demonstrate a realtime, smooth, performance of an industrial robot subject to dynamic obstacles.

This paper introduces the task of Planar Robot Casting (PRC): where one planar motion of a robot arm holding one end of a cable causes the other end to slide across the plane toward a desired target. PRC allows the cable to reach points beyond the robot workspace and has applications for cable management in homes, warehouses, and factories. To efficiently learn a PRC policy for a given cable, we propose Real2Sim2Real, a self-supervised framework that automatically collects physical trajectory examples to tune parameters of a dynamics simulator using Differential Evolution, generates many simulated examples, and then learns a policy using a weighted combination of simulated and physical data. We evaluate Real2Sim2Real with three simulators, Isaac Gym-segmented, Isaac Gym-hybrid, and PyBullet, two function approximators, Gaussian Processes and Neural Networks (NNs), and three cables with differing stiffness, torsion, and friction. Results with 240 physical trials suggest that the PRC policies can attain median error distance (as % of cable length) ranging from 8 % to 14 %, outperforming baselines and policies trained on only real or only simulated examples. Code, data, and videos are available at https://tinyurl.com/robotcast.

Dynamic models play a key role in robot motion generation and control and the identification of inertial parameters is a critical component for obtaining an accurate dynamic model of a robot. This paper presents a novel iterative primitive shape division method for the inertia parameter identification of floating-base robots. Describing a robot by a set of primitive shapes with uniform mass distributions, the method iteratively divides the primitive shapes into smaller ones and refines their masses, which quickly converges to yielding the true inertia parameters of the robot. This method guarantees the physical consistency of the obtained parameters, possesses a high computational efficiency for online deployment, and works without contact force measurement. Furthermore, it can be used to estimate the position and magnitude of an external load applied to the robot. Simulations and experiments on a quadruped robot have been conducted to verify the effectiveness and efficiency of the proposed method.

We propose a novel and pragmatic framework for traffic scene perception with roadside cameras. The proposed framework covers a full-stack of roadside perception pipeline for infrastructure-assisted autonomous driving, including object detection, object localization, object tracking, and multi-camera information fusion. Unlike previous vision-based perception frameworks rely upon depth offset or 3D annotation at training, we adopt a modular decoupling design and introduce a landmark-based 3D localization method, where the detection and localization can be well decoupled so that the model can be easily trained based on only 2D annotations. The proposed framework applies to either optical or thermal cameras with pinhole or fish-eye lenses. Our framework is deployed at a two-lane roundabout located at Ellsworth Rd. and State St., Ann Arbor, MI, USA, providing 7×247\times 24 real-time traffic flow monitoring and high-precision vehicle trajectory extraction. The whole system runs efficiently on a low-power edge computing device with all-component end-to-end delay of less than 20ms.

Real-time kinodynamic trajectory planning in dy-namic environments is critical yet challenging for autonomous driving. In this paper, we propose an efficient trajectory plan-ning system for autonomous driving in complex dynamic sce-narios through iterative and incremental path-speed optimization. Exploiting the decoupled structure of the planning prob-lem, a path planner based on Gaussian process first generates a continuous arc-length parameterized path in the Frenét frame, considering static obstacle avoidance and curvature constraints. We theoretically prove that it is a good generalization of the well-known jerk optimal solution. An efficient s-t graph search method is introduced to find a speed profile along the generated path to deal with dynamic environments. Finally, the path and speed are optimized incrementally and iteratively to ensure kinodynamic feasibility. Various simulated scenarios with both static obstacles and dynamic agents verify the effectiveness and robustness of our proposed method. Experimental results show that our method can run at 20 Hz. The source code is released as an open-source package.

We propose a hybrid model predictive control algorithm, consensus complementarity control (C3), for systems that make and break contact with their environment. Many state-of-the-art controllers for tasks which require initiating contact with the environment, such as locomotion and manipulation, require a priori mode schedules or are so computationally complex that they cannot run at real-time rates. We present a method, based on the alternating direction method of multipliers (ADMM), capable of high-speed reasoning over potential contact events. Via a consensus formulation, our approach enables parallelization of the contact scheduling problem. We validate our results on three numerical examples, including two frictional contact problems, and physical experimentation on an underactuated multi-contact system.

Flexible-joint manipulators are governed by complex nonlinear dynamics, defining a challenging control problem. In this work, we propose an approach to learn an outer-loop joint trajectory tracking controller with deep reinforcement learning. The controller represented by a stochastic policy is learned in under two hours directly on the real robot. This is achieved through bounded reference correction actions and use of a model-free off-policy learning method. In addition, an informed policy initialization is proposed, where the agent is pre-trained in a learned simulation. We test our approach on the 7 DOF manipulator of a Baxter robot. We demonstrate that the proposed method is capable of consistent learning across multiple runs when applied directly on the real robot. Our method yields a policy which significantly improves the trajectory tracking accuracy in comparison to the vendor-provided controller, generalizing to an unseen payload.

In this paper, we examine the problem of push recovery for bipedal robot locomotion and present a reactive decision-making and robust planning framework for locomotion resilient to external perturbations. Rejecting perturbations is an essential capability of bipedal robots and has been widely studied in the locomotion literature. However, adversarial disturbances and aggressive turning can lead to negative lateral step width (i.e., crossed-leg scenarios) with unstable motions and self-collision risks. These motion planning problems are computationally difficult and have not been explored under a hierarchically integrated task and motion planning method. We explore a planning and decision-making framework that closely ties linear-temporal-logic-based reactive synthesis with trajectory optimization incorporating the robot's full-body dynamics, kinematics, and leg collision avoidance constraints. Between the high-level discrete symbolic decision-making and the low-level continuous motion planning, behavior trees serve as a reactive interface to handle perturbations occurring at any time of the locomotion process. Our experimental results show the efficacy of our method in generating resilient recovery behaviors in response to diverse perturbations from any direction with bounded magnitudes.

In this paper we address mobile manipulation planning problems in the presence of sensing and environmental uncertainty. In particular, we consider mobile sensing manipulators operating in environments with unknown geometry and uncertain movable objects, while being responsible for accomplishing tasks requiring grasping and releasing objects in a logical fashion. Existing algorithms either do not scale well or neglect sensing and/or environmental uncertainty. To face these challenges, we propose a hybrid control architecture, where a symbolic controller generates high-level manipulation commands (e.g., grasp an object) based on environmental feedback, an informative planner designs paths to actively decrease the uncertainty of objects of interest, and a continuous reactive controller tracks the sparse waypoints comprising the informative paths while avoiding a priori unknown obstacles. The overall architecture can handle environmental and sensing uncertainty online, as the robot explores its workspace. Using numerical simulations, we show that the proposed architecture can handle tasks of increased complexity while responding to unanticipated adverse configurations.

ReachBot is a new concept for planetary exploration, consisting of a small body and long, lightweight extending arms loaded primarily in tension. The arms are equipped with spined grippers for anchoring on rock surfaces. The design and testing of a planar prototype is presented here. Experiments with rock grasping and coordinated locomotion illustrate the advantages of low inertia passive grippers, triggered by impact and using stored mechanical energy for the internal force. Gripper design involves a trade-off among the range of possible grasp angles, maximum grasp force, required triggering force, and required reset force. The current prototype can pull with up to 8N when gripping volcanic rock, limited only by the strength of the 3D printed components. Calculations predict a maximum pull of 26N for the same spines and stronger materials.