IROS 2023 论文列表

The air-sea interface is vital in studying heat and energy exchange between the sea and air. The field observation technology of the air-sea interface is an effective way to explore the nature of the air-sea interface. This paper presents an observation system called DANDELION for the air-sea interface environment. The system includes an automatic surface vehicle (ASV), a launching device using a pair of high-speed rotating friction wheels, and eight dandelion-like micro profilers. This system can implement the environmental observation of the air-sea interface in an extensive range through micro profilers' rapid and multi-point placement. The DANDELION system was characterized by establishing the friction wheel launching mechanism model and summarizing the effects of different wings on the profiler performance. A series of experiments were conducted in Qiandao Lake, China, to characterize the DANDELION system. We demonstrate the developed system with data from field experiments, which show very high flexibility and feasibility to observe the air-sea interface, implying potential applications in ocean transient phenomena observation.

DAMON leverages manifold learning and variational autoencoding to achieve obstacle avoidance, allowing for motion planning through adaptive graph traversal in a pre-learned low-dimensional hierarchically-structured manifold graph that captures intricate motion dynamics between a robotic arm and its obstacles. This versatile and reusable approach is applicable to various collaboration scenarios. The primary advantage of DAMON is its ability to embed information in a low-dimensional graph, eliminating the need for repeated computation required by current sampling-based methods. As a result, it offers faster and more efficient motion planning with significantly lower computational overhead and memory footprint. In summary, DAMON is a breakthrough methodology that addresses the challenge of dynamic obstacle avoidance in robotic systems and offers a promising solution for safe and efficient human-robot collaboration. Our approach has been experimentally validated on a 7-DoF robotic manipulator in both simulation and physical settings. DAMON enables the robot to learn and generate skills for avoiding previously-unseen obstacles while achieving predefined objectives. We also optimize DAMON's design parameters and performance using an analytical framework. Our approach outperforms mainstream methodologies, including RRT, RRT*, Dynamic RRT*, L2RRT, and MpNet, with 40% more trajectory smoothness and over 65% improved latency performance, on average.

Robot grippers are widely used in industrial automation for pick-and-place tasks on a variety of objects. Whilst the majority of commercial grippers are capable of establishing stable grasps, few can perform in-hand-manipulation (IHM). IHM is has the potential to increase robotic motion efficiency, yet most IHM-capable manipulation platforms are anthropomorphic in nature and cost over 10,000, posing a barrier to entry for many. In this work we propose a IHM capable gripper platform that is open-source and may be assembled for £150 (10,000, posing a barrier to entry for many. In this work we propose a IHM capable gripper platform that is open-source and may be assembled for £150 (162) and access to a 3D printer. The gripper consists of two fully actuated 2DOF fingers, each of which is based on a five-bar linkage mechanism with one link extended. The fingers are modular, allowing the gripper to be easily expanded into 3+ finger configurations via simple modification of the central mount. We define the inverse kinematics and effective workspace of the gripper (via the use of Freudenstein equations), providing guidance for translation and rotation of gripped objects. We demonstrate the gripper's ability to manipulate a 1-inch cube's pose within a ±5% error margin and rotate various other YCB objects via open-loop position control.

Visual servoing represents a control strategy capable of driving dynamical systems from the current to the desired pose, when the only available information is the images generated at both poses. In this work, we analyze vulnerability of such systems and introduce two types of attacks to deceive visual servoing controller within a wheeled mobile robotic system. The attack goal is to alter the visual servoing procedure in such a way that mobile robot achieves the pose defined by an attacker instead of the desired one. Specifically, the attacks exploit image transformations developed using a methodology based on simulated annealing. The main difference between the attacks is the considered threat model - i.e., how the attacker has infiltrated the system. The first attack assumes the real-time camera feed has been compromised and thus, the images from the current pose are modified (e.g., during the acquisition or communication); for the second, only the desired destination image is potentially altered. Finally, in 3D simulations and real- world experiments, we show the effectiveness of cyber-attacks.

We present CurveQuad, a miniature curved origami quadruped that is able to self-fold and unfold, crawl, and steer, all using a single actuator. CurveQuad is designed for planar manufacturing, with parts that attach and stack sequentially on a flat body. The design uses 4 curved creases pulled by 2 pairs of tendons from opposite ends of a link on a 270°servo. It is 8 cm in the longest direction and weighs 10.9 g. Rotating the horn pulls the tendons inwards to induce folding. Continuing to rotate the horn shears the robot, enabling the robot to shuffle forward while turning in either direction. We experimentally validate the robot's ability to fold, steer, and unfold by changing the magnitude of horn rotation. We also demonstrate basic feedback control by steering towards a light source from a variety of starting positions and orientations, and swarm aggregation by having 4 robots simultaneously steer towards the light. The results demonstrate the potential of using curved crease origami in self-assembling and deployable robots with complex motions such as locomotion.

Addressing domain shifts for complex perception tasks in autonomous driving has long been a challenging problem. In this paper, we show that existing domain adaptation methods pay little attention to the content mismatch issue between source and target domains, thus weakening the domain adaptation per-formance and the decoupling of domain-invariant and domain-specific representations. To solve the aforementioned problems, we propose an image-level domain adaptation framework that aims at adapting source-domain images to the target domain with content-aligned source-target image pairs. Our framework consists of three mutually beneficial modules in a cycle: a cross-domain content alignment module to generate source-target pairs with consistent content representations in a self-supervised manner, a reference-guided image synthesis based on the generated content-aligned source-target image pairs, and a contrastive learning module to self-supervise domain-invariant feature extractor. Our contrastive appearance adaptation is task-agnostic and robust to complex perception tasks in autonomous driving. Our proposed method demonstrates state-of-the-art results in cross-domain object detection, semantic segmentation, and depth estimation as well as better image synthesis ability qualitatively and quantitatively.

We present a reinforcement learning (RL) framework that enables quadrupedal robots to perform soccer goalkeeping tasks in the real world. Soccer goalkeeping with quadrupeds is a challenging problem, that combines highly dynamic locomotion with precise and fast non-prehensile object (ball) manipulation. The robot needs to react to and intercept a potentially flying ball using dynamic locomotion maneuvers in a very short amount of time, usually less than one second. In this paper, we propose to address this problem using a hierarchical model-free RL framework. The first component of the framework contains multiple control policies for distinct locomotion skills, which can be used to cover different regions of the goal. Each control policy enables the robot to track random parametric end-effector trajectories while performing one specific locomotion skill, such as jump, dive, and sidestep. These skills are then utilized by the second part of the framework which is a high-level planner to determine a desired skill and end-effector trajectory in order to intercept a ball flying to different regions of the goal. We deploy the proposed framework on a Mini Cheetah quadrupedal robot and demonstrate the effectiveness of our framework for various agile interceptions of a fast-moving ball in the real world.

In this work, we consider a team of heterogeneous robots equipped with various types and quantities of resources, and tasked with supplying these resources to multiple dynamic demand locations. We present an adaptive control policy that enables robots to serve a dynamic demand: we allow demand to deplete as robots supply resources, and we allow demand injection and movement of demand locations. We show that the demand is input-to-state stable (ISS) under our proposed resource dynamics, and thus the robots can drive the demand to a steady state. Finally, we present simulations and hardware experiments to demonstrate our approach, and demonstrate the benefits of coverage over a persistent monitoring approach.

We address the problem of multi-goal multi robot path planning (MG-MRPP) via counterexample guided abstraction refinement (CEGAR) framework. MG-MRPP generalizes the standard discrete multi-robot path planning (MRPP) problem. While the task in MRPP is to navigate robots in an undirected graph from their starting vertices to one individual goal vertex per robot, MG-MRPP assigns each robot multiple goal vertices and the task is to visit each of them at least once. Solving MG-MRPP not only requires finding collision free paths for individual robots but also determining the order of visiting robot's goal vertices so that common objectives like the sum-of-costs are optimized. We use the Boolean satisfiability (SAT) techniques as the underlying paradigm. A specifically novel in this work is the use of non-refined abstractions when formulating the MG-MRPP problem as SAT. While the standard CEGAR approach for MG-MRPP does not encode collision elimination constraints in the initial abstraction and leave them to subsequent refinements. The novel CEGAR approach leaves some abstractions deliberately non-refined. This adds the necessity to post-process the answers obtained from the underlying SAT solver as these answers slightly differ from the correct MG-MRPP solutions. Non-refining however yields order-of-magnitude smaller SAT encodings than those of the previous CEGAR approach and speeds up the overall solving process.

We present a coordination method for multiple mobile manipulators to sort objects in clutter. We consider the object rearrangement problem in which the objects must be sorted into different groups in a particular order. In clutter, the order constraints could not be easily satisfied since some objects occlude other objects so the occluded ones are not directly accessible to the robots. Those objects occluding others need to be moved more than once to make the occluded objects accessible. Such rearrangement problems fall into the class of nonmonotone rearrangement problems which are computation-ally intractable. While the nonmonotone problems with order constraints are harder, involving with multiple robots requires another computation for task allocation. In this work, we aim to develop a fast method, albeit sub-optimally, for the multi-robot coordination for ordered sorting in clutter. The proposed method finds a sequence of objects to be sorted using a search such that the order constraint in each group is satisfied. The search can solve nonmonotone instances that require temporal relocation of some objects to access the next object to be sorted. Once a complete sorting sequence is found, the objects in the sequence are assigned to multiple mobile manipulators using a greedy task allocation method. We develop four versions of the method with different search strategies. In the experiments, we show that our method can find a sorting sequence quickly (e.g., 4.6 sec with 20 objects sorted into five groups) even though the solved instances include hard nonmonotone ones. The extensive tests and the experiments in simulation show the ability of the method to solve the real-world sorting problem using multiple mobile manipulators.

Biological agents, such as humans and animals, are capable of making decisions out of a very large number of choices in a limited time. They can do so because they use their prior knowledge to find a solution that is not necessarily optimal but good enough for the given task. In this work, we study the motion coordination of multiple drones under the above-mentioned paradigm, Bounded Rationality (BR), to achieve cooperative motion planning tasks. Specifically, we design a prior policy that provides useful goal-directed navigation heuristics in familiar environments and is adaptive in unfamiliar ones via Reinforcement Learning augmented with an environment-dependent exploration noise. Integrating this prior policy in the game-theoretic bounded rationality framework allows agents to quickly make decisions in a group considering other agents' computational constraints. Our investigation assures that agents with a well-informed prior policy increase the efficiency of the collective decision-making capability of the group. We have conducted rigorous experiments in simulation and in the real world to demonstrate that the ability of informed agents to navigate to the goal safely can guide the group to coordinate efficiently under the BR framework.

Cooperative Simultaneous Localization and Mapping (C-SLAM) is an active research topic in mobile robotics. However, its application in the field of autonomous driving is rare. While the advent of Vehicle-to-Everything (V2X) communication has empowered Connected Autonomous Vehicles (CAV) to exchange data with each other, recent research on CAV cooperation tasks has primarily focused on cooperative perception and global positioning improvement. Techniques for organizing multiple CAV to work together to achieve localization and mapping in unknown environments have not been actively explored. We propose a C-SLAM system for CAVs that employs sparse LiDAR feature representations to enable vehicles to exchange data using standard V2X messages. The system was tested in real environments using two connected vehicles. The results show that the proposed V2X-based C-SLAM system can operate in both centralized and decentralized manners and output accurate pose estimates and global maps, showing promising application possibilities.

This paper presents a new control scheme for cooperative dual-arm robots manipulating heavy objects. The proposed method uses the full dynamical model of the kinematically coupled robot system and builds on a hierarchical quadratic programming (HQP) formulation to enforce dynamical inequality constraints such as joint torques or internal loads. This ensures optimal tracking of an object trajectory, while additional objectives with lower priority are optimized on the prior solution space. Therefore, the redundancy of the inherent load distribution problem between the two arms can be eliminated. With this approach, higher object loads can be manipulated compared to non-optimized methods. Simulations with a 14 degree of freedom (dof) dual-arm robotic system demonstrate the effectiveness of the proposed control method. The real-time feasibility is guaranteed with an average computation time of less than 0.35 milliseconds at a control rate of 1 kilohertz.

Dense packing in pick-and-place systems is an important feature in many warehouse and logistics applications. Prior work in this space has largely focused on planning algorithms in simulation, but real-world packing performance is often bottlenecked by the difficulty of perceiving 3D object geometry in highly occluded, partially observed scenes. In this work, we present a fully-convolutional shape completion model, F-CON, which can be easily combined with off-the-shelf planning methods for dense packing in the real world. We also release a simulated dataset, COB-3D-v2, that can be used to train shape completion models for real-word robotics applications, and use it to demonstrate that F-CON outperforms other state-of-the-art shape completion methods. Finally, we equip a real-world pick-and-place system with F-CON, and demonstrate dense packing of complex, unseen objects in cluttered scenes. Across multiple planning methods, F-CON enables substantially better dense packing than other shape completion methods.

In recent years, depth sensors have become more and more affordable and have found their way into a growing amount of robotic systems. However, mono- or multi-modal sensor registration, often a necessary step for further pro-cessing, faces many challenges on raw depth images or point clouds. This paper presents a method of converting depth data into images capable of visualizing spatial details that are basically hidden in traditional depth images. After noise removal, a neighborhood of points forms two normal vectors whose difference is encoded into this new conversion. Compared to Bearing Angle images, our method yields brighter, higher-contrast images with more visible contours and more details. We tested feature-based pose estimation of both conversions in a visual odometry task and RGB-D SLAM. For all tested features, AKAZE, ORB, SIFT, and SURF, our new Flexion images yield better results than Bearing Angle images and show great potential to bridge the gap between depth data and classical computer vision. Source code is available here: https://rlsch.github.io/depth-flexion-conversion.

One of the primary benefits of emerging powered prosthetic legs is their ability to facilitate step-over-step stair ascent by providing positive mechanical work. Existing control methods typically have distinct steady-state activity modes for walking and stair ascent, where activity transitions involve discretely switching between controllers and often must be initiated with a particular leg. However, these discrete transitions do not necessarily replicate able-bodied joint biomechanics, which have been shown to continuously adjust over a transition stride. This paper presents a phase-based kinematic controller for a powered knee-ankle prosthesis that enables continuous, biomimetic transitions between walking and stair ascent. The controller tracks joint angles from a data-driven kinematic model that continuously interpolates between the steady-state kinematic models, and it allows both the prosthetic and intact leg to lead the transitions. Results from experiments with two transfemoral amputee participants indicate that knee and ankle kinematics smoothly transition between walking and stair ascent, with comparable or lower root mean square errors compared to variations from able-bodied data.

In this work, we exploit the recent notion of closed-loop state sensitivity to critically compare three typical controllers for a quadrotor UAV with the goal of evaluating the impact of controller choice, gain tuning and shape of the reference trajectory in minimizing the sensitivity of the closed-loop system against uncertainties in the model parameters. To this end, we propose a novel optimization problem that takes into account both the shape of the reference trajectory and the controller gains. We then run a large statistical campaign for comparing the performance of the three controllers which provides some interesting insight for the goal of increasing closed-loop robustness against parametric uncertainties.

This work focuses on the capability of Mobile Manipulators to effectively control and maneuver cart-like non-holonomic systems. These cart-like systems are passive-wheeled objects with nonholonomic constraints with varying inertial parameters. We derive the dynamic equations of the cart-like system using a constrained Euler-Lagrange formulation and propose a Linear Quadratic Regulator controller to move the cart along a desired trajectory using external forces (applied by the MM) at a given contact point. For the MM, we present a control architecture to i) control the mobile base to keep the cart inside the workspace of the manipulator and ii) a control Lyapunov function formulation to control the manipulator in torque control, while decoupling the motion of the base from the arm and applying the required wrench onto the object. We validate our approach experimentally, using a MM to push a shopping cart and track desired trajectories. These experiments show the accuracy of the control architecture to track the desired trajectories for carts with different inertial parameters and improve the controllability of the system by changing the contact point on the cart.

Learning long-horizon tasks such as navigation has presented difficult challenges for successfully applying reinforcement learning to robotics. From another perspective, under known environments, sampling-based planning can robustly find collision-free paths in environments without learning. In this work, we propose Control Transformer that models return-conditioned sequences from low-level policies guided by a sampling-based Probabilistic Roadmap (PRM) planner. We demonstrate that our framework can solve long-horizon navigation tasks using only local information. We evaluate our approach on partially-observed maze navigation with MuJoCo robots, including Ant, Point, and Humanoid. We show that Control Transformer can successfully navigate through mazes and transfer to unknown environments. Additionally, we apply our method to a differential drive robot (Turtlebot3) and show zero-shot sim2real transfer under noisy observations.