检索结果-内蒙古大学图书馆

Curriculum-based Sensing Reduction in Simulation to Real-World Transfer for In-hand Manipulation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Tao, Lingfeng Zhang, Jiucai Zheng, Qiaojie Zhang, Xiaoli Oklahoma State University 563 Engineering North StillwaterOK74075 United States GAC R&D Center Silicon Valley SunnyvaleCA94085 United States Colorado School of Mines Intelligent Robotics and Systems Lab 1500 Illinois St GoldenCO80401 United States

Simulation to Real-World Transfer allows affordable and fast training of learning-based robots for manipulation tasks using Deep Reinforcement Learning methods. Currently, Sim2Real uses Asymmetric Actor-Critic approaches to reduce the rich idealized features in simulation to the accessible ones in the real world. However, the feature reduction from the simulation to the real world is conducted through an empirically defined one-step curtail. Small feature reduction does not sufficiently remove the actor’s features, which may still cause difficulty setting up the physical system, while large feature reduction may cause difficulty and inefficiency in training. To address this issue, we proposed Curriculum-based Sensing Reduction to enable the actor to start with the same rich feature space as the critic and then get rid of the hard-to-extract features step-by-step for higher training performance and better adaptation for real-world feature space. The reduced features are replaced with random signals from a Deep Random Generator to remove the dependency between the output and the removed features and avoid creating new dependencies. The methods are evaluated on the Allegro robot hand in a real-world in-hand manipulation task. The results show that our methods have faster training and higher task performance than baselines and can solve real-world tasks when selected tactile features are reduced. Copyright © 2023, The Authors. All rights reserved.

关键词： Curricula

Design and Development of the FlexBE WebUI with Introductory Tutorials

学校读者我要写书评

暂无评论

Journal of Computing Sciences in Colleges

Journal of Computing Sciences in Colleges 2024年第3期40卷

作者： Samuel Raymond Grace Walters Josh Luzier David C. Conner Capable Humanitarian Robotics and Intelligent Systems Lab Department of Physics Computer Science and Engineering Christopher Newport University Newport News VA

Hierarchical Finite State Machines (HFSMs) continue to be a popular strategy for high-level behavioral control in robotics. Within the Robot Operating System (ROS) ecosystem, the Flexible Behavior Engine (FlexBE) provides an easy to use Python implementation with an intuitive user interface for runtime monitoring and "collaborative autonomy." In this paper, we describe the latest evolution of the FlexBE UI, that leverages FastAPI to provide a Python-based native ROS node interface, to a web-based user interface. Additionally, the paper describes a "quick-start" introductory tutorial to state machines using FlexBE. This system provides a natural introduction to HFSM-based techniques.

关键词：

Stable In-hand Manipulation with Finger Specific Multi-agent Shadow Reward

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Tao, Lingfeng Zhang, Jiucai Zhang, Xiaoli Oklahoma State University 563 Engineering North StillwaterOK74075 United States Colorado School of Mines Intelligent Robotics and Systems Lab 1500 Illinois St GoldenCO80401 United States The GAC R&D Center Silicon Valley SunnyvaleCA94085 United States

Deep Reinforcement Learning has shown its capability to solve the high degrees of freedom in control and the complex interaction with the object in the multi-finger dexterous in-hand manipulation tasks. Current DRL approaches prefer sparse rewards to dense rewards for the ease of training but lack behavior constraints during the manipulation process, leading to aggressive and unstable policies that are insufficient for safety-critical in-hand manipulation tasks. Dense rewards can regulate the policy to learn stable manipulation behaviors with continuous reward constraints but are hard to empirically define and slow to converge optimally. This work proposes the Finger-specific Multi-agent Shadow Reward (FMSR) method to determine the stable manipulation constraints in the form of dense reward based on the state-action occupancy measure, a general utility of DRL that is approximated during the learning process. Information Sharing (IS) across neighboring agents enables consensus training to accelerate the convergence. The methods are evaluated in two in-hand manipulation tasks on the Shadow Hand. The results show FMSR+IS converges faster in training, achieving a higher task success rate and better manipulation stability than conventional dense reward. The comparison indicates FMSR+IS achieves a comparable success rate even with the behavior constraint but much better manipulation stability than the policy trained with a sparse reward. Copyright © 2023, The Authors. All rights reserved.

关键词： Degrees of freedom (mechanics)

A Novel Multi-Stream Informer Used for Lower Extremity Joint Angle Estimation 12

学校读者我要写书评

暂无评论

A Novel Multi-Stream Informer Used for Lower Extremity Joint...

12th International Conference on CYBER Technology in Automation, Control, and intelligent systems, CYBER 2022

作者： Zhou, Xin Zhang, Liming Liu, Jiaqing Ye, Jiancong Wang, Can Wu, Xinyu Guangdong Provincial Key Lab of Robotics and Intelligent System Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen518055 China Cas Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institutes of Advanced Technology China University of Science and Technology of China China University of Chinese Academy of Sciences China South China University of Technology China Department of Mechanical and Automation Engineering The Chinese University of Hong Kong Hong Kong

ISBN: (数字)9781665472678

ISBN: (纸本)9781665472678

In lower extremity exoskeleton rehabilitation systems, synergy and proportionality between the human lower extremity and the exoskeleton robot have been a critical goal to pursue. In recent years, changeable deep learning models based on multi-source signals have been used for lower limb continuous motion estimation. During the process, multimode signals are fed into the time processing model as time series. However, there are some deficiencies in this framework, on the one hand, what kind of multi-source combination can be more efficient;on the other hand, most of the existing models need several iterations or network stacking to complete the extraction of temporal information. This study set out to address the above problems, we put forward Multi-Stream Informer (MS-Informer), using Multi-Stream combination and sparse attention mechanisms for lower limb continuous motion estimation. In the experiment, we compare MS-Informer and Temporal Convolutional Network (TCN) and Long Short-term Memory (LSTM) under distinctive multi-source combinations. In summary, MS-informer performs well in multiple evaluations compared with other models. Experimental results illustrate that the MS-Informer is more efficacious in lower limb joint angle estimation and is a promising method for achieving coordinated control of lower limb exoskeletons. © 2022 IEEE.

关键词： Motion estimation

Multi-Robot System for Autonomous Cooperative Counter-UAS Missions: Design, Integration, and Field Testing

学校读者我要写书评

暂无评论

Multi-Robot System for Autonomous Cooperative Counter-UAS Mi...

IEEE International Workshop on Safety, Security, and Rescue robotics (SSRR)

作者： Antonella Barisic Marlan Ball Noah Jackson Riley McCarthy Nasib Naimi Luca Strässle Jonathan Becker Maurice Brunner Julius Fricke Lovro Markovic Isaac Seslar David Novick Jonathan Salton Roland Siegwart Stjepan Bogdan Rafael Fierro Laboratory for Robotics and Intelligent Control Systems (LAR-ICS) Faculty of Electrical Engineering and Computing University of Zagreb Zagreb Croatia Electrical and Computer Engineering Department University of New Mexico Albuquerque NM USA ETH Zurich Autonomous Systems Lab (ASL) Zurich Switzerland Sandia National Laboratories Albuquerque New Mexico USA

ISBN: (纸本)9781665456814

With the rapid development of technology and the proliferation of uncrewed aerial systems (UAS), there is an immediate need for security solutions. Toward this end, we propose the use of a multi-robot system for autonomous and cooperative counter-UAS missions. In this paper, we present the design of the hardware and software components of different complementary robotic platforms: a mobile uncrewed ground vehicle (UGV) equipped with a LiDAR sensor, an uncrewed aerial vehicle (UAV) with a gimbal-mounted stereo camera for air-to-air inspections, and a UAV with a capture mechanism equipped with radars and camera. Our proposed system features 1) scalability to larger areas due to the distributed approach and online processing, 2) long-term cooperative missions, and 3) complementary multimodal perception for the detection of multirotor UAVs. In field experiments, we demonstrate the integration of all subsystems in accomplishing a counter-UAS task within an unstructured environment. The obtained results confirm the promising direction of using multi-robot and multi-modal systems for C-UAS.

关键词： Scalability Robot vision systems Autonomous aerial vehicles Cameras Software Safety Security

Tactile grasp stability classification based on graph convolutional networks

学校读者我要写书评

暂无评论

Tactile grasp stability classification based on graph convol...

2021 IEEE International Conference on Real-Time Computing and robotics, RCAR 2021

作者： Mi, Tingting Que, Dashun Fang, Senlin Zhou, Zhenning Ye, Chaoxiang Liu, Chengliang Yi, Zhengkun Wu, Xinyu School of Information Engineering Wuhan University of Technology Wuhan430070 China Shenzhen Institute of Advanced Technology CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen518055 China Guangdong Provincial Key Lab of Robotics and Intelligent System Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen518055 China Shenzhen Key Laboratory of Smart Sensing and Intelligent Systems Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen518055 China SIAT Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen518055 China

ISBN: (纸本)9781665436786

One of the challenges for robots to grasp unknown objects is to predict whether objects will fall at the beginning of grasping. Evaluating robotic grasp state accurately and efficiently is a significant step to address this issue. In this paper, based on the different fusion approaches of multi-sensor tactile signals, we propose two novel methods based on Graph Convolution Network (GCN) for robotic stability classification. Specifically, we propose two deep learning methods including GCN based on data-level fusion (GCN-DF) and GCN based on feature-level fusion (GCN-FF). We explore the optimal parameters for transforming sensor signals into a graph structure. Furthermore, we verify the effectiveness of the proposed methods on the BioTac Grasp Stability (BiGS) dataset. The experimental results prove that the proposed approaches achieve higher classification accuracy than Support Vector Machine (SVM) and Long Short-Term Memory (LSTM). © 2021 IEEE.

关键词： Convolution

How to Model Brushless Electric Motors for the Design of Lightweight Robotic systems

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lee, Ung Hee Shepherd, Tor Kim, Sangbae De, Avik Su, Hao Gregg, Robert Mooney, Luke Rouse, Elliott Neurobionics Lab. University of Michigan Ann ArborMI48109 United States Department of Mechanical Engineering University of Michigan United States The Robotics Institute University of Michigan United States Department of Mechanical Engineering Massachusetts Institute of Technology CambridgeMA02139 United States Department of Electrical and Systems Engineering University of Pennsylvania PhiladelphiaPA19104 United States Lab of Biomechatronics and Intelligent Robotics Department of Mechanical Engineering and Aerospace Engineering North Carolina State University RaleighNC27695 United States Department of Electrical Engineering and Computer Science University of Michigan United States Dephy Inc. MaynardMA01754 United States

A key step in the development of lightweight, high performance robotic systems is the modeling and selection of permanent magnet brushless direct current (BLDC) electric motors. Typical modeling analyses are completed a priori, and provide insight for properly sizing a motor for an application, specifying the required operating voltage and current, as well as assessing the thermal response and other design attributes e.g. transmission ratio). However, to perform these modeling analyses, proper information about the motor’s characteristics are needed, which are often obtained from manufacturer datasheets. Through our own experience and communications with manufacturers, we have noticed a lack of clarity and standardization in modeling BLDC motors, compounded by vague or inconsistent terminology used in motor datasheets. The purpose of this tutorial is to concisely describe the governing equations for BLDC motor analyses used in the design process, as well as highlight potential errors that can arise from incorrect usage. We present a power-invariant conversion from phase and line-to-line reference frames to a familiar q-axis DC motor representation, which provides a "brushed" analogue of a three phase BLDC motor that is convenient for analysis and design. We highlight potential errors including incorrect calculations of winding resistive heat loss, improper estimation of motor torque via the motor’s torque constant, and incorrect estimation of the required bus voltage or resulting angular velocity limitations. A unified and condensed set of governing equations is available for designers in Appendix A. The intent of this work is to provide a consolidated mathematical foundation for modeling BLDC motors that addresses existing confusion and fosters high performance designs of future robotic systems. © 2023, CC BY-NC-ND.

关键词： Design

Preliminary study of online real-time control system for lower extremity exoskeletons based on EEG and sEMG fusion

学校读者我要写书评

暂无评论

Preliminary study of online real-time control system for low...

IEEE International Conference on robotics and Biomimetics

作者： Wenju Li Keyong Shao Chi Zhu Yue Ma Wujing Cao Meng Yin Lijun Yang Mingxiang Luo Xinyu Wu School of Electrical Engineering Information Northeast Petroleum University Daqing China Guangdong Provincial Key Lab of Robotics and Intelligent System Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen Guangdong China Department of Systems Life Engineering Maebashi Institute of Technology Meabashi Japan

ISBN: (纸本)9781665481106

Motion imaging (MI) is widely used in exoskele-ton robot control direction. However, online real-time con-trol of exoskeletons based on biological signals is difficult to achieve. Because the anti-interference of biological signals in real life is poor and affected by individual differences, the control of auxiliary equipment by the realization of human-machine interface is still facing great challenges. In this paper, a human-machine interface (HMI) based on hybrid biological signals is designed and online real-time control of lower extremity exoskeletons is realized. Based on a comprehensive analysis of human walking, the HMI uses the electromyography signals (upper extremities) and EEG signals generated during real exercise as input signals. Different analysis methods are designed for different modal signals. We extracted four features of sEMG signals to distinguish between different movement processes in the human body. For EEG signals, the use of Independent Component Analysis (ICA) increases the robustness of the system. Event-related desynchronization/synchronization (ERS/ERD) and power spectral densities are reused for analysis from the perspective of the interaction of multiple information in the channel, time and frequency domains. The human-machine interface integrates EEG and EMG multimodal biological signals, and the average recognition accuracy of different subjects has reached 93.7%. In order to increase the recognition accuracy of practical applications and reduce the differences between different individuals, a window voting mechanism is designed. Healthy subjects were tested, and the recognition accuracy of the lower limb exoskeleton online real-time assistance system reached 99%, with high accuracy and reliability.

关键词： Legged locomotion Exoskeletons Feature extraction Real-time systems Electroencephalography Biology Robustness

On the Use of Torque Measurement in Centroidal State Estimation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Khorshidi, Shahram Gazar, Ahmad Rotella, Nicholas Naveau, Maximilien Righetti, Ludovic Bennewitz, Maren Khadiv, Majid Max-Planck Institute for Intelligent Systems Tübingen Germany Agility Robotics Oregon United States LAAS CNRS Toulouse France Tandon School of Engineering New York University United States Humanoid Robots Lab University of Bonn Germany

State-of-the-art legged robots are either capable of measuring torque at the output of their drive systems, or have transparent drive systems which enable the computation of joint torques from motor currents. In either case, this sensor modality is seldom used in state estimation. In this paper, we propose to use joint torque measurements to estimate the centroidal states of legged robots. To do so, we project the whole-body dynamics of a legged robot into the nullspace of the contact constraints, allowing expression of the dynamics independent of the contact forces. Using the constrained dynamics and the centroidal momentum matrix, we are able to directly relate joint torques and centroidal states dynamics. Using the resulting model as the process model of an Extended Kalman Filter (EKF), we fuse the torque measurement in the centroidal state estimation problem. Through real-world experiments on a quadruped robot executing different gaits, we demonstrate that the estimated centroidal states from our torque-based EKF drastically improve the recovery of these quantities compared to direct computation. © 2022, CC BY.

关键词： State estimation