IROS 2021 论文列表

As more robots are being deployed into human environments, a human-aware navigation planner needs to handle multiple contexts that occur in indoor and outdoor environments. In this paper, we propose a tunable human-aware robot navigation planner that can handle a variety of human-robot contexts. We present the architecture of the system and discuss the features along with some implementation details. Then we present a detailed analysis of various simulated human-robot contexts using the proposed planner. Further, we show that our system performs better when compared with an exiting human-aware planner in various contexts. Finally, we show the results in a real-world scenario after deploying our system on a real robot.

Human physical assistance requires the assistant to tune both his trajectory and impedance in order to assist an individual as well as be guided by him. In this study we propose a controller for teleoperated human assistance that allows the assistant to guide the assisting robot in both trajectory and impedance. We propose to use the inherent perturbations in the task, induced by the elderly or stroke patient, for impedance estimation, while a simple neuroscience based filter allows the reference estimation of the operator. We tested our impedance estimation and the controller as a whole in two experiments in which a human operator guided a robot suffering force perturbations that simulated a human patient.

Research in biomechanics hypothesizes that human motion is optimal with respect to an unknown cost function that varies depending on the action and/or task. This unknown cost function is often approximated as the weighted sum of a set of features or basis cost functions. As a person performs a sequence of actions, the weights associated to each of these basis functions are likely to vary over time. Given a human demonstration and the corresponding cost weight trajectory recovered via inverse optimal control (IOC), this paper proposes an optimal control (OC) method that can generate robot motion based on human movement using time-varying cost function weights. By using time-varying weights, the proposed optimal control method can handle changing optimization criteria without segmentation. The method is evaluated both in simulation and with recorded human data. Using human demonstration data, we demonstrate the reproduction of pick-and-place motions with an average end-effector error at the pick place location within 0.82 cm, which is significantly lower than the average trajectory error, indicating that the approach correctly prioritizes reaching the pick and place locations without manual segmentation.

Technologies for estimating self-position and orientation are important for both humans and robots. These technologies allow robots to perform tasks such as carrying objects and allow people to reach their destinations. Although self-position estimation technologies using GPS and laser rangefinders have been developed, few methods can be used by both humans and robots. Therefore, we developed a method that can estimate three-dimensional position and orientation using visual markers and an inertia measurement unit (IMU). Self-position can be measured with high accuracy by using a visual marker and monocular camera, but such measurement data is discrete and sparse. In contrast, an IMU can continuously measure acceleration data, but data obtained from an acceleration sensor are double-integrated, which increases position error. By combining visual marker and IMU information, position error calculations based on the acceleration sensor can be corrected, and the movement path of the object can be estimated. In demonstration experiments, the proposed method accurately estimates the three-dimensional movement distance when a person walks about 13 m, with an average error of about 40.3mm.

Quadruped robots possess advantages on different terrains over other types of mobile robots by virtue of their flexible choices of foothold points. It is crucial to integrate terrain perception with motion planning to exploit the potential of quadruped robots. We propose a novel hierarchical terrain-aware control (HTC) framework, which leverages deep reinforcement learning (DRL) for the high-level planner and optimal control for the low-level controller. In general, traditional control methods yield better stability by using an optimization algorithm. In addition, DRL is able to offer more adaptive behavior. Our approach makes full use of the advantages of these two methods and possesses better adaptability and stability in challenging natural environments. Furthermore, the global height map of the terrain serves as visual information for the DRL, which determines the desired footholds for the robot’s leg swings and body postures. Optimal control calculates the torque of the joints on the standing legs to maintain body balance. Our method is tested on various terrains both simulated and real environments. The experimental results show that HTC can effectively enhance the adaptability of the quadruped robot by coordinating body posture.

This paper presents a hierarchical segment-based optimization method for Simultaneous Localization and Mapping (SLAM) system. First we propose a reliable trajectory segmentation method that can be used to increase efficiency in the back-end optimization. Then we propose a buffer mechanism for the first time to improve the robustness of the segmentation. During the optimization, we use global information to optimize the frames with large error, and interpolation instead of optimization to update well-estimated frames to hierarchically allocate the amount of computation according to error of each frame. Comparative experiments on the benchmark show that our method greatly improves the efficiency of optimization with almost no drop in accuracy, and outperforms existing high-efficiency optimization method by a large margin.

Leader-follower navigation is a popular class of multi-robot algorithms where a leader robot leads the follower robots in a team. The leader has specialized capabilities or mission critical information (e.g. goal location) that the followers lack, and this makes the leader crucial for the mission’s success. However, this also makes the leader a vulnerability -an external adversary who wishes to sabotage the robot team’s mission can simply harm the leader and the whole robot team’s mission would be compromised. Since robot motion generated by traditional leader-follower navigation algorithms can reveal the identity of the leader, we propose a defense mechanism of hiding the leader’s identity by ensuring the leader moves in a way that behaviorally camouflages it with the followers, making it difficult for an adversary to identify the leader. To achieve this, we combine Multi-Agent Reinforcement Learning, Graph Neural Networks and adversarial training. Our approach enables the multi-robot team to optimize the primary task performance with leader motion similar to follower motion, behaviorally camouflaging it with the followers. Our algorithm outperforms existing work that tries to hide the leader’s identity in a multi-robot team by tuning traditional leader-follower control parameters with Classical Genetic Algorithms. We also evaluated human performance in inferring the leader’s identity and found that humans had lower accuracy when the robot team used our proposed navigation algorithm.

Ceiling mobile robots are anticipated for trans-porting component parts in production sites. In our previous study, we had developed a crawler-type ceiling mobile robot named "HanGrawler." In this study, we aim to realize 1 m/s and 90°/s speed movement on par with existing ground carriers, and reveal the factors contributing to the improvement in speed and stabilization. Accordingly, we develop HanGrawler 2, which inherits the basic approach of HanGrawler. In HanGrawler 2, the mechanism for hanging from the ceiling moves faster, and the driving power of the linear and rotational movements is improved drastically. Newly installed sensors detect the hanging mechanism and robot body position relative to the ceiling plate. By calculating the insertion timing based on the position of the hanging mechanism, the certainty of insertion is improved. Support rollers pressed against the ceiling stabilize the traveling posture of HanGrawler 2. Performance evaluation experiments confirm that HanGrawler 2 can travel linearly at a speed of 1 m/s with a 40 kg load and turn at a rotational speed of 90°/s. Based on the experimental results, it is confirmed that the design solutions for HanGrawler 2 are effective in realizing reliable ceiling mobility and that the high-responsiveness inherited from the original HanGrawler mechanism contributes to the realization of high-speed movement.

This paper introduces HOPPY, an open-source, low-cost, robust, and modular kit for robotics education. The robot dynamically hops around a rotating gantry with a fixed base. The kit is intended to lower the entry barrier for studying dynamic robots and legged locomotion with real systems. It bridges the theoretical content of fundamental robotic courses with real dynamic robots by facilitating and guiding the software and hardware integration. This paper describes the topics which can be studied using the kit, lists its components, discusses preferred practices for implementation, presents results from experiments with the simulator and the real system, and suggests further improvements. A simple heuristic-based controller is described to achieve velocities up to 1.7m/s, navigate small objects, and mitigate external disturbances when the robot is aided by a counterweight. HOPPY was utilized as the subject of a semester-long project for the Robot Dynamics and Control course at the University of Illinois at Urbana-Champaign. The positive feedback from the students and instructors about the hands-on activities during the course motivates us to share this kit and continue improving it in the future.

The use of reinforcement learning (RL) has led to huge advancements in the field of robotics. However data scarcity, brittle convergence and the gap between simulation & real world environments, mean that most common RL approaches are subject to over fitting and fail to generalise to unseen environments. Hardware agnostic policies would mitigate this by allowing a single network to operate in a variety of test domains, where dynamics vary due to changes in robotic morphologies or internal parameters. We utilise the idea that learning to adapt a known and successful control policy is easier and more flexible than jointly learning numerous control policies for different morphologies.This paper presents the idea of Hardware Agnostic Reinforcement Learning using Adversarial selection (HARL-A). In this approach training examples are sampled using a novel adversarial loss function. This is designed to self regulate morphologies based on their learning potential. Simply applying our learning potential based loss function to current state-of-the-art already provides ~ 30% improvement in performance. Meanwhile experiments using the full implementation of HARL-A report an average increase of 70% to a standard RL baseline and 55% compared with current state-of-the-art.

Virtually all robot control methods benefit from the availability of an accurate mathematical model of the robot. However, obtaining a sufficient amount of informative data for constructing dynamic models can be difficult, especially when the models are to be learned during robot deployment. Under such circumstances, standard data-driven model learning techniques often yield models that do not comply with the physics of the robot. We extend a symbolic regression algorithm based on Single Node Genetic Programming by including the prior model information into the model construction process. In this way, symbolic regression automatically builds models that compensate for theoretical or empirical model deficiencies. We experimentally demonstrate the approach on two real-world systems: the TurtleBot 2 mobile robot and the Parrot Bebop 2 drone. The results show that the proposed model-learning algorithm produces realistic models that fit well the training data even when using small training sets. Passing the prior model information to the algorithm significantly improves the model accuracy while speeding up the search.

This paper studies the group multi-object tracking (MOT) problem in dynamic pedestrian environments, with intended application to safe navigation for autonomous vehicles. We complete a full autonomous vehicle navigation pipeline from object detection, tracking, grouping, to risk map generation and safe path planning. Our main contribution is to instantiate a group multi-object tracking algorithm, which provides the crucial grouped activity information, i.e. group position, group velocity, group size, to the risk map generator, and therewith produce a stable and robust risk map for the downstream safe path planner. Experimental results with real world data show the socially acceptable, robust and stable performance of the proposed algorithm over its individual MOT counterpart.

Humans have a natural ability to effortlessly comprehend linguistic commands such as “park next to the yellow sedan” and instinctively know which region of the road the vehicle should navigate. Extending this ability to autonomous vehicles is the next step towards creating fully autonomous agents that respond and act according to human commands. To this end, we propose the novel task of Referring Navigable Regions (RNR), i.e., grounding regions of interest for navigation based on the linguistic command. RNR is different from Referring Image Segmentation (RIS), which focuses on grounding an object referred to by the natural language expression instead of grounding a navigable region. For example, for a command “park next to the yellow sedan,” RIS will aim to segment the referred sedan, and RNR aims to segment the suggested parking region on the road. We introduce a new dataset, Talk2Car-RegSeg, which extends the existing Talk2car [1] dataset with segmentation masks for the regions described by the linguistic commands. A separate test split with concise manoeuvre-oriented commands is provided to assess the practicality of our dataset. We benchmark the proposed dataset using a novel transformer-based architecture. We present extensive ablations and show superior performance over baselines on multiple evaluation metrics. A downstream path planner generating trajectories based on RNR outputs confirms the efficacy of the proposed framework.

We address the problem of robot localization using ground penetrating radar (GPR) sensors. Current approaches for localization with GPR sensors require a priori maps of the system’s environment as well as access to approximate global positioning (GPS) during operation. In this paper, we propose a novel, real-time GPR-based localization system for unknown and GPS-denied environments. We model the localization problem as an inference over a factor graph. Our approach combines 1D single-channel GPR measurements to form 2D image submaps. To use these GPR images in the graph, we need sensor models that can map noisy, high-dimensional image measurements into the state space. These are challenging to obtain a priori since image generation has a complex dependency on subsurface composition and radar physics, which itself varies with sensors and variations in subsurface electromagnetic properties. Our key idea is to instead learn relative sensor models directly from GPR data that map non-sequential GPR image pairs to relative robot motion. These models are incorporated as factors within the factor graph with relative motion predictions correcting for accumulated drift in the position estimates. We demonstrate our approach over datasets collected across multiple locations using a custom designed experimental rig. We show reliable, real-time localization using only GPR and odometry measurements for varying trajectories in three distinct GPS-denied environments.

Many cities suffer from a shortage of parking spaces. Research in high density parking (HDP) focuses on how to increase the capacity of parking lots by allowing vehicles to block each other but temporarily give way to other vehicles by driving autonomously upon request. Previous works on HDP did not consider mixing different parking strategies and ignored the possibility of gridlock when multiple vehicles move simultaneously. In this paper, we describe the design of autonomous parking lots, which allows the deployment of different parking strategies in different regions in a parking lot. We present algorithms for checking whether adding a vehicle to an autonomous parking lot can lead to gridlock. Our simulation shows that autonomous parking lots can hold 60% more vehicles given the same amount of space.

Kinematic and force synergies can be used to reduce the complexity and dimensionality of the motion generation and control problem, as well as facilitate the mechanical implementation of robotic hands. In this paper we present a novel implementation of hardware synergies realized on the actuation level by leveraging a novel reconfigurable electric actuation topology principle. The proposed electric actuation topology enables different actuation synergies by changing the interconnections among the actuators at the electrical/motor driver level. We describe the synergies and their implementation in a port-based context, and elaborate how equivalent hard and soft synergies emerge from the electric power flow within different actuation topologies. We realize the reconfigurable electric actuation topology scheme in the HERI III hand, a novel robust and powerful robotic gripper, also introduced in this paper, resulting in easy to control behaviours like on industrial grippers or underactuated hands, but with the high grasping versatility of fully actuated hands. Finally we present grasping experiments performed on the HERI III hand that clearly show the desired behaviours and validate the innovative proposed scheme.

Grasp pose detection generates the position and orientation of the robot end-effector to grasp objects from the RGB or RGB-D image. In this paper, we propose a novel grasp pose detection network that generates 3-DOF grasp poses using the RGB image. The network follows the anchor-based object detection pipeline and incorporates the angle detection unit. Furthermore, we redesign the grasp angle predictor with a classification unit to increase the accuracy of grasp pose rotation estimation. Our method classifies the prediction angle densely in contrast with the previous regression method or sparse classification method. Moreover, an angle smooth label is designed to avoid the sudden change of the angle regression loss caused by the periodic property of the angle. We validate our algorithm on Cornell Grasp Dataset and obtain a higher detection accuracy than the state-of-the-art method. The real scenario experiment also proves the effectiveness of our method. The robot equipped with the parallel gripper achieves a 96.4% grasp success rate.

Capturing scene dynamics and predicting the future scene state is challenging but essential for robotic manipulation tasks, especially when the scene contains both rigid and deformable objects. In this work, we contribute a simulation environment and generate a novel dataset for task-specific manipulation, involving interactions between rigid objects and a deformable bag. The dataset incorporates a rich variety of scenarios including different object sizes, object numbers and manipulation actions. We approach dynamics learning by proposing an object-centric graph representation and two modules which are Active Prediction Module (APM) and Position Prediction Module (PPM) based on graph neural networks with an encode-process-decode architecture. At the inference stage, we build a two-stage model based on the learned modules for single time step prediction. We combine modules with different prediction horizons into a mixed-horizon model which addresses long-term prediction. In an ablation study, we show the benefits of the two-stage model for single time step prediction and the effectiveness of the mixed-horizon model for long-term prediction tasks. Supplementary material is available at https://github.com/wengzehang/deformable_rigid_interaction_prediction