检索结果-内蒙古大学图书馆

Learning to Model Diverse Driving Behaviors in Highly Interactive autonomous Driving Scenarios with Multi-Agent Reinforcement Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Weiwei, Liu Wenxuan, Hu Wei, Jing Lanxin, Lei Lingping, Gao Yong, Liu The Advanced Perception on Robotics and Intelligent Learning Lab College of Control Science and Enginneering Zhejiang University Hangzhou310027 China The Advanced Perception on Robotics and Intelligent Learning Lab Huzhou Institute Zhejiang University Huzhou China College of Information Engineering Huzhou University Huzhou China Department of Autonomous Driving Lab Alibaba DAMO Academy Hangzhou China

autonomous vehicles trained through Multi-Agent Reinforcement Learning (MARL) have shown impressive results in many driving scenarios. However, the performance of these trained policies can be impacted when faced with diverse driving styles and personalities, particularly in highly interactive situations. This is because conventional MARL algorithms usually operate under the assumption of fully cooperative behavior among all agents and focus on maximizing team rewards during training. To address this issue, we introduce the Personality Modeling Network (PeMN), which includes a cooperation value function and personality parameters to model the varied interactions in high-interactive scenarios. The PeMN also enables the training of a background traffic flow with diverse behaviors, thereby improving the performance and generalization of the ego vehicle. Our extensive experimental studies, which incorporate different personality parameters in high-interactive driving scenarios, demonstrate that the personality parameters effectively model diverse driving styles and that policies trained with PeMN demonstrate better generalization compared to traditional MARL methods. Copyright © 2024, The Authors. All rights reserved.

关键词： Reinforcement learning

FlowMP: Learning Motion Fields for Robot Planning with Conditional Flow Matching

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Nguyen, Khang Le, An T. Pham, Tien Huber, Manfred Peters, Jan Vu, Minh Nhat Learning and Adaptive Robotics Lab University of Texas Arlington United States Intelligent Autonomous Systems Lab TU Darmstadt Germany Cognitive Robotics Lab University of Manchester United Kingdom SAIROL Darmstadt Germany Hessian.AI Darmstadt Germany TU Wien Vienna Austria GmbH Vienna Austria

Prior flow matching methods in robotics have primarily learned velocity fields to morph one distribution of trajectories into another. In this work, we extend flow matching to capture second-order trajectory dynamics, incorporating acceleration effects either explicitly in the model or implicitly through the learning objective. Unlike diffusion models, which rely on a noisy forward process and iterative denoising steps, flow matching trains a continuous transformation (flow) that directly maps a simple prior distribution to the target trajectory distribution without any denoising procedure. By modeling trajectories with second-order dynamics, our approach ensures that generated robot motions are smooth and physically executable, avoiding the jerky or dynamically infeasible trajectories that first-order models might produce. We empirically demonstrate that this second-order conditional flow matching yields superior performance on motion planning benchmarks, achieving smoother trajectories and higher success rates than baseline planners. These findings highlight the advantage of learning acceleration-aware motion fields, as our method outperforms existing motion planning methods in terms of trajectory quality and planning success. Copyright © 2025, The Authors. All rights reserved.

关键词： Trajectories

Robust Sliding Mode Based Finite-time Bilateral Shared Teleoperation System with Unsymmetrical Time-Varying Delay* 48

学校读者我要写书评

暂无评论

Robust Sliding Mode Based Finite-time Bilateral Shared Teleo...

48th Annual Conference of the IEEE Industrial Electronics Society, IECON 2022

作者： Islam, Shafiqul Sunda-Meya, Anderson Xavier University of Louisiana Dual Degree Engineering Program Intelligent Robotics Mechatronics & Autonomous Systems Lab New Orleans70125 United States Xavier University of Louisiana College of Arts & Sciences New Orleans70125 United States

ISBN: (数字)9781665480253

ISBN: (纸本)9781665480253

This paper presents sliding mode-based finite-time synchronization algorithms for bilateral shared teleoperation systems under unsymmetrical time-varying delay and uncertainty. The synchronization algorithms are designed by using Lyapunov and sliding mode control theory. First, the model-based finite-time synchronization algorithms are designed by using sliding mode control theory with the presence of time-varying delays and uncertainty. Then, robust adaptive finite-time synchronization algorithms are designed for bilateral shared teleoperation systems with the presence of time-varying delays. Adaptive learning algorithms learn and adapt with uncertainty associated with the unmodeled dynamics and external disturbances. Lyapunov analysis shows that the tracking errors between master and slave manipulators converge to the sliding surface in finite time. It also shows that the tracking errors asymptotically converge to zero on the sliding surface. The design provides finite-time convergence of the states to reach the sliding surface as opposed to asymptotic convergence-based bilateral shared teleoperation systems. The design and analysis do not use the exact bound of the uncertainty. In contrast with asymptotic design, finite-time convergence can ensure faster and more robust tracking in the presence of uncertainty and time-varying delays. Unlike reported designs, the proposed algorithms can ensure convergence with the presence of unsymmetrical time-varying delays. © 2022 IEEE.

关键词： Time delay

Enabling Service Robots to Open Self-Closing Doors using Deep RL and Generative Models

学校读者我要写书评

暂无评论

Research Square

Research Square 2025年

作者： Sun, Yufeng Zhang, Lin Ma, Ou Intelligent Robotics and Autonomous Systems Lab University of Cincinnati CincinnatiOH45221 United States Department of Physics and Astronomy University of Central Arkansas 201 Donaghey Avenue ConwayAR72035 United States

Various mobile service robots have been applied in human-centered indoor environments such as hospitals, schools, workshops, and office buildings for tasks like food or material delivery, surveillance, cleaning, and disinfection. However, they still require human assistance to access rooms with doors, particularly self-closing doors. In this paper, we propose a deep reinforcement learning (DRL)-based solution to enhance existing service robots, enabling them to autonomously open self-closing doors using a simple and inexpensive assistive device. This device is utilized to unlatch doors through the integration of the latest object detection and sensor technologies. Focusing on pulling open self-closing doors, we train an end-to-end control policy in a simplified simulation environment and deploy it to a real-world robot. Additionally, generative models, a variational autoenoder (VAE) and a cycle-consistent adversarial network (CycleGAN), are incorporated into both the training and deployment stages to improve the robustness of the door-opening control policy. The proposed solution for opening self-closing doors was validated in both simulation and real-world experiments. © 2025, CC BY.

关键词： Reinforcement learning

FSG-Net: a Deep Learning model for Semantic Robot Grasping through Few-Shot Learning

学校读者我要写书评

暂无评论

FSG-Net: a Deep Learning model for Semantic Robot Grasping t...

IEEE International Conference on robotics and Automation (ICRA)

作者： Leonardo Barcellona Alberto Bacchin Alberto Gottardi Emanuele Menegatti Stefano Ghidoni Department of Information Engineering Intelligent Autonomous System Lab University of Padova Padua Italy Politecnico di Torino Torino Italy IT +Robotics srl Vicenza Italy

Robot grasping has been widely studied in the last decade. Recently, Deep Learning made possible to achieve remarkable results in grasp pose estimation, using depth and RGB images. However, only few works consider the choice of the object to grasp. Moreover, they require a huge amount of data for generalizing to unseen object categories. For this reason, we introduce the Few-shot Semantic Grasping task where the objective is inferring a correct grasp given only five labelled images of a target unseen object. We propose a new deep learning architecture able to solve the aforementioned problem, leveraging on a Few-shot Semantic Segmentation module. We have evaluated the proposed model both in the Graspnet dataset and in a real scenario. In Graspnet, we achieve 40,95% accuracy in the Few-shot Semantic Grasping task, outperforming baseline approaches. In the real experiments, the results confirmed the generalization ability of the network.

关键词：

Convergence Analysis for Delay Dependent LFC Scheme for Power Systems Networks Over Open Communication Networks with Disturbances

学校读者我要写书评

暂无评论

Convergence Analysis for Delay Dependent LFC Scheme for Powe...

IEEE International Conference on Industrial Technology (ICIT)

作者： Shafiqul Islam Dual-Degree Computer Science and Engineering Intelligent Robotics Mechatronics and Autonomous Systems Lab NCF 263 Xavier University of Louisiana New Orleans LA USA

This paper presents a delay-dependent convergence criterion for load frequency control (LFC) for smart power grid networks with the presence of disturbances and time-varying delays. The LFC scheme comprises delayed remote measurement and control signals with local state feedback signals. The convergence criteria are developed by using Lyapunov-Krasovskii functional for both symmetrical and asymmetrical time-varying delays associated with remote measurement and control channel. First, the analysis shows that all the power system states are bounded and asymptotically stable without the presence of disturbances. Then, the convergence analysis proves that all the power systems states are bounded and their bounds guarantee the convergence property of the $H_{\infty}$ control with the presence of bounded disturbances. These convergence properties are achieved by using well know Linear Matrix Inequalities (LMIs). Compared with the existing design, the proposed analysis uses both symmetrical and asymmetrical time-varying delays in the measurement channel from RTU to the control center and in the control channel from the control center to the power generation unit. In contrast with reported designs, the proposed design and analysis uses the disturbances associated with the closed loop systems appearing from modeling errors and other external load disturbances and faults.

关键词：

A 3-Step Optimization Framework with Hybrid Models for a Humanoid Robot's Jump Motion

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Qi, Haoxiang Yu, Zhangguo Chen, Xuechao Liu, Yaliang Yi, Chuanku Dong, Chencheng Meng, Fei Huang, Qiang The Intelligent Robotics Institute School of Mechatronical Engineering Beijing Institute of Technology Beijing100081 China The International Joint Research Laboratory of Biomimetic Robots and Systems Ministry of Education Beijing100081 China The National Key Lab of Autonomous Intelligent Unmanned Systems Beijing100081 China

High dynamic jump motions are challenging tasks for humanoid robots to achieve environment adaptation and obstacle crossing. The trajectory optimization is a practical method to achieve high-dynamic and explosive jumping. This paper proposes a 3-step trajectory optimization framework for generating a jump motion for a humanoid robot. To improve iteration speed and achieve ideal performance, the framework comprises three sub-optimizations. The first optimization incorporates momentum, inertia, and center of pressure (CoP), treating the robot as a static reaction momentum pendulum (SRMP) model to generate corresponding trajectories. The second optimization maps these trajectories to joint space using effective Quadratic Programming (QP) solvers. Finally, the third optimization generates whole-body joint trajectories utilizing trajectories generated by previous parts. With the combined consideration of momentum and inertia, the robot achieves agile forward jump motions. A simulation and experiments (Fig. 1) of forward jump with a distance of 1.0 m and 0.5 m height are presented in this paper, validating the applicability of the proposed framework. Note to Practitioners-The motivation of this paper stems from the need to improve jumping performance of humanoid robots. By comprehensively considering factors such as robot posture, centroidal angular momentum, and landing foot placement, the algorithm enhances the robot's ability to navigate complex environments. This capability is crucial for applications that require overcoming obstacles, such as in search and rescue or inspection tasks. Improved jumping ability can significantly boost environmental adaptability, allowing robots to perform effectively in diverse conditions, and it also represents an exploration of the high-dynamic motion capabilities of humanoid robots. Future research will focus on integrating visual and perceptual information to enhance decision-making. Copyright © 2025, The Authors. All rights reserve

关键词： Anthropomorphic robots

autonomous Forest Inventory with Legged Robots: System Design and Field Deployment

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Mattamala, Matias Chebrolu, Nived Casseau, Benoit Freißmuth, Leonard Frey, Jonas Tuna, Turcan Hutter, Marco Fallon, Maurice Oxford Robotics Institute The University of Oxford United Kingdom Robotic Systems Lab ETH Zurich Switzerland Technical University of Munich Germany Autonomous Learning Group Max Planck Institute for Intelligent Systems Germany

We present a solution for autonomous forest inventory with a legged robotic platform. Compared to their wheeled and aerial counterparts, legged platforms offer an attractive balance of endurance and low soil impact for forest applications. In this paper, we present the complete system architecture of our forest inventory solution which includes state estimation, navigation, mission planning, and real-time tree segmentation and trait estimation. We present preliminary results for three campaigns in forests in Finland and the UK and summarize the main outcomes, lessons, and challenges. Our UK experiment at the Forest of Dean with the ANYmal D legged platform, achieved an autonomous survey of a 0.96 ha plot in 20 min, identifying over 100 trees with typical DBH accuracy of 2 cm. © 2024, CC BY.

关键词： Machine design

A Comparison of Imitation Learning Algorithms for Bimanual Manipulation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Drolet, Michael Stepputtis, Simon Kailas, Siva Jain, Ajinkya Peters, Jan Schaal, Stefan Amor, Heni Ben The School of Computing and Augmented Intelligence Arizona State University Tempe United States The Robotics Institute Carnegie Mellon University Pittsburgh United States Intrinsic Mountain View United States The Intelligent Autonomous Systems Lab. TU Darmstadt Darmstadt Germany

Amidst the wide popularity of imitation learning algorithms in robotics, their properties regarding hyperparameter sensitivity, ease of training, data efficiency, and performance have not been well-studied in high-precision industry-inspired environments. In this work, we demonstrate the limitations and benefits of prominent imitation learning approaches and analyze their capabilities regarding these properties. We evaluate each algorithm on a complex bimanual manipulation task involving an over-constrained dynamics system in a setting involving multiple contacts between the manipulated object and the environment. While we find that imitation learning is well suited to solve such complex tasks, not all algorithms are equal in terms of handling environmental and hyperparameter perturbations, training requirements, performance, and ease of use. We investigate the empirical influence of these key characteristics by employing a carefully designed experimental procedure and learning environment. © 2024, CC BY.

关键词： Contrastive Learning