检索结果-内蒙古大学图书馆

TALKER: A Task-Activated Language Model Based Knowledge-Extension Reasoning System

IEEE robotics AND AUTOMATION LETTERS 2025年第2期10卷 1026-1033页

作者： Lou, Jiabin Shi, Rongye Lin, Yuxin Wang, Qunbo Wu, Wenjun Beihang Univ Beijing 100191 Peoples R China

Training drones to execute complex collective tasks via multi-agent reinforcement learning presents significant challenges. To address these challenges, this letter introduces the Task-Activated Language model-based Knowledge-Extension Reasoning system. Specifically, we trained drones in two fine-grained skills and developed an action primitive library based on these capabilities, enabling a hierarchical approach to managing complex swarm operations. Leveraging this primitive library, we employ large language models to perform task planning, continuously refining the planning outcomes based on external user feedback. Successful task codes are temporarily stored within the action primitive library, with their utilization being authorized based on internal feedback from maintainers. We defined this process as knowledge expansion. In addition, more refined customized prompts are generated based on task descriptions and the action primitive documentation, a mechanism referred to as Task Activation. Our system synergistically integrates task activation and knowledge expansion mechanisms, enabling continuous evolution through human feedback to effectively manage extensive swarms in the execution of complex collective tasks. Experimental results demonstrate the superior performance of our system in various drone swarm tasks, including collaborative search, object tracking, cooperative interception, and aerial patrol.

关键词： Drones Planning Navigation Cognition Libraries Robots Robot kinematics Reinforcement learning Training Resource management ai-enabled robotics autonomous systems large language models reinforcement learning task planning

来源：评论

学校读者我要写书评

暂无评论

Modular Reinforcement Learning for a Quadrotor UAV With Decoupled Yaw Control

引用

IEEE robotics AND AUTOMATION LETTERS 2025年第1期10卷 572-579页

作者： Yu, Beomyeol Lee, Taeyoung George Washington Univ Flight Dynam & Control Lab Mech & Aerosp Engn Washington DC 20051 USA

This letter presents modular reinforcement learning (RL) frameworks for the low-level control of a quadrotor, with direct control of yawing motion. While traditional monolithic RL approaches have been successfully applied to real-world autonomous flight, they often struggle to precisely control both translational and yawing motions due to their distinct dynamic characteristics and coupling. Moreover, training a large-scale monolithic network typically requires extensive training data to achieve broad generalization. To address these issues, we decompose the quadrotor dynamics into translational and yaw subsystems and assign a dedicated modular RL agent to each. This design significantly improves performance, as each RL agent is trained for its specific purpose and integrated in a synergistic way. It further enhances robustness, as potential failures within one module have minimal impact on the other, promoting fault tolerance. These improvements are demonstrated through flight experiments achieved via zero-shot sim-to-real transfer, where it is shown that the proposed modular policies substantially enhance training efficiency, tracking performance, and adaptability to real-world conditions.

关键词： Quadrotors Reinforcement learning Training Robustness Mixers Autonomous aerial vehicles Aerodynamics Torque Rotors Motors Aerial systems: Mechanics and control ai-enabled robotics reinforcement learning (RL)

来源：评论

学校读者我要写书评

暂无评论

Probabilistically Correct Language-Based Multi-Robot Planning Using Conformal Prediction

引用

IEEE robotics AND AUTOMATION LETTERS 2025年第1期10卷 160-167页

作者： Wang, Jun He, Guocheng Kantaros, Yiannis Washington Univ St Louis (WashU) Preston M Green Dept Elect & Syst Engn St Louis MO 63130 USA Vanderbilt Univ Dept Comp Sci Nashville TN 37240 USA

This paper addresses task planning problems for language-instructed robot teams. Tasks are expressed in natural language (NL), requiring the robots to apply their skills at various locations and semantic objects. Several recent works have addressed similar planning problems by leveraging pre-trained Large Language Models (LLMs) to design effective multi-robot plans. However, these approaches lack performance guarantees. To address this challenge, we introduce a new distributed LLM-based planner, called S-ATLAS for Safe plAnning for Teams of Language-instructed AgentS, that can achieve user-defined mission success rates. This is accomplished by leveraging conformal prediction (CP), a distribution-free uncertainty quantification tool. CP allows the proposed multi-robot planner to reason about its inherent uncertainty, due to imperfections of LLMs, in a distributed fashion, enabling robots to make local decisions when they are sufficiently confident and seek help otherwise. We show, both theoretically and empirically, that the proposed planner can achieve user-specified task success rates, assuming successful plan execution, while minimizing the average number of help requests. We provide comparative experiments against related works showing that our method is significantly more computational efficient and achieves lower help rates.

关键词： Multi-robot systems task and motion planning ai-enabled robotics planning under uncertainty Multi-robot systems task and motion planning ai-enabled robotics planning under uncertainty

来源：评论

学校读者我要写书评

暂无评论

Make a Donut: Hierarchical EMD-Space Planning for Zero-Shot Deformable Manipulation With Tools

引用

IEEE robotics AND AUTOMATION LETTERS 2025年第4期10卷 3270-3277页

作者： You, Yang Shen, William B. Deng, Congyue Geng, Haoran Wei, Songlin Wang, He Guibas, Leonidas Stanford Univ Dept Comp Sci Stanford CA 94305 USA Peking Univ CFCS Beijing 100871 Peoples R China

Deformable object manipulation stands as one of the most captivating yet formidable challenges in robotics. While previous techniques have predominantly relied on learning latent dynamics through demonstrations, typically represented as either particles or images, there exists a pertinent limitation: acquiring suitable demonstrations, especially for long-horizon tasks, can be elusive. Moreover, basing learning entirely on demonstrations can hamper the model's ability to generalize beyond the demonstrated tasks. In this work, we introduce a demonstration-free hierarchical planning approach capable of tackling intricate long-horizon deformable manipulation tasks without necessitating any training. We employ large language models (LLMs) to articulate a high-level, stage-by-stage plan corresponding to a specified task. For every individual stage, the LLM provides both the tool's name and the Python code to craft intermediate subgoal point clouds. With the tool and subgoal for a particular stage at our disposal, we present a granular closed-loop model predictive control strategy. This leverages Differentiable Physics with Point-to-Point correspondence (DiffPhysics-P2P) loss in the Earth Mover Distance (EMD) space, applied iteratively. Experimental findings affirm that our technique surpasses multiple benchmarks in dough manipulation, spanning both short and long horizons. Remarkably, our model demonstrates robust generalization capabilities to novel and previously unencountered complex tasks without any preliminary demonstrations. We further substantiate our approach with experimental trials on real-world robotic platforms.

关键词： Planning Point cloud compression Robots Codes Physics Training Aerospace electronics Python Pins Cognition Task and motion planning manipulation planning perception for grasping and manipulation ai-enabled robotics vision language model application

来源：评论

学校读者我要写书评

暂无评论

LPM-Net: A Data-Driven Resource-Efficient Predictive Motion Planner for Mobile Robots

引用

NEURAL PROCESSING LETTERS 2025年第1期57卷 1-15页

作者： Amirhosseini, Fakhreddin Nilforoushan, Zahra Mirtaheri, Seyedeh Leili Kharazmi Univ Elect & Comp Engn Dept Tehran Iran Univ Calabria Dept Informat Modeling Elect & Syst Engn DIMES Arcavacata Di Rende Italy

A data-driven predictive motion planner for mobile robots, referred to as LPM-Net, has been proposed in this paper. Conventional predictive motion planners are computationally expensive, often resulting in insufficient throughput on mobile robot hardware. LPM-Net is an imitation learning-assisted local predictive non-holonomic motion planner that is capable of learning from conventional motion planners regarded as paradigm models and replicating their behavior while satisfying the same kinodynamic constraints. In addition, LPM-Net is compatible with GPU and TPU hardware, allowing for faster and more efficient processing. LPM-Net uses convolutional and recurrent long short-term memory deep neural networks to predict steering commands. This has improved computational efficiency which allows autonomous vehicles to be equipped with more cost-effective computers. In the present study, LPM-Net was tuned to mimic the behavior of a model predictive controller paradigm model. Measurements in this study demonstrate that the proposed mimic planner, LPM-Net, consumes approximately half the processing power of the conventional predictive planner, albeit with a slight increase in hesitation when reaching goals.

关键词： ai-enabled robotics Imitation learning Non-holonomic motion planning Model predictive controller

来源：评论

学校读者我要写书评

暂无评论

Dynamics as Prompts: In-Context Learning for Sim-to-Real System Identifications

引用

IEEE robotics AND AUTOMATION LETTERS 2025年第4期10卷 3190-3197页

作者： Zhang, Xilun Liu, Shiqi Huang, Peide Han, William Jongwon Lyu, Yiqi Xu, Mengdi Zhao, Ding Carnegie Mellon Univ Dept Mech Engn Pittsburgh PA 15213 USA

Sim-to-real transfer remains a significant challenge in robotics due to the discrepancies between simulated and real-world dynamics. Traditional methods like Domain Randomization often fail to capture fine-grained dynamics, limiting their effectiveness for precise control tasks. In this work, we propose a novel approach that dynamically adjusts simulation environment parameters online using in-context learning. By leveraging past interaction histories as context, our method adapts the simulation environment dynamics to match real-world dynamics without requiring gradient updates, resulting in faster and more accurate alignment between simulated and real-world performance. We validate our approach across two tasks: object scooping and table air hockey. In the sim-to-sim evaluations, our method significantly outperforms the baselines on environment parameter estimation by 80% and 42% in the object scooping and table air hockey setups, respectively. Furthermore, our method achieves at least 70% success rate in sim-to-real transfer on object scooping across three different objects. By incorporating historical interaction data, our approach delivers efficient and smooth system identification, advancing the deployment of robots in dynamic real-world scenarios.

关键词： Machine learning for robot control ai-enabled robotics reinforcement learning sim-to-real transfer Machine learning for robot control ai-enabled robotics reinforcement learning sim-to-real transfer

来源：评论

学校读者我要写书评

暂无评论

Task Decomposition and Self-Evaluation Mechanisms for Home Healthcare Robots Using Large Language Models

引用

IEEE ACCESS 2025年 13卷 65726-65736页

作者： Liu, Liteng Zhang, Sen Jiang, Yangmin Guo, Jingzhen Zhao, Wenlong Univ Shanghai Sci & Technol Hlth Sci & Engn Shanghai 200093 Peoples R China Shanghai Univ Med & Hlth Sci Ctr Collaborat Res Shanghai 201318 Peoples R China Shanghai Jiao Tong Univ China Hosp Dev Inst Res Ctr Med Intelligent Dev Shanghai 200025 Peoples R China

A system leveraging Large Language Models (LLMs), is proposed to address the limitations of current models primarily used for conversational purposes. While user interactions are excelled by ChatGPT, instability and safety issues are encountered when generating control codes for manipulator. This study focuses on the field of home healthcare services and proposes a new approach to control robots equipped with self-developed dexterous hand. The objective of this paper is to develop and evaluate a new system, Elderly Care GPT, aimed at improving the control of robots for home healthcare services. The system utilizes the Chain-of-Thought (CoT) prompting technique to enhance the accuracy of model-generated control codes. In conjunction with a self-evaluation mechanism, it ensures that the control codes undergo an accuracy evaluation before execution, thereby ensuring safety. Additionally, the system decomposes tasks into basic groups to improve the robot's task execution capabilities. The performance of the model is evaluated across eight real-world scenarios, covering classic home care tasks. In real-world evaluations, Elderly Care GPT is compared with LLaMA3 and LLaMA3+ Prompt Words, and a comparison is made with mainstream models used for controlling robots on the market. The results show that Elderly Care GPT achieves a task success rate of up to 97.3%, demonstrating exceptional performance. This paper concludes that Elderly Care GPT outperforms current models used for robot control in home healthcare services, achieving high task success rates and effectively handling complex tasks. The use of Chain-of-Thought prompting, a self-evaluation mechanism, and task decomposition significantly contributes to the model's success.

关键词： Robots Codes Older adults Chatbots Accuracy Large language models Manipulators Data models Robot control Prompt engineering ai-enabled robotics manipulation planning domestic robotics large language model

来源：评论

学校读者我要写书评

暂无评论

Planning and Reasoning with 3D Deformable Objects for Hierarchical Text-to-3D Robotic Shaping

引用

IEEE robotics and Automation Letters 2025年第6期10卷 6215-6222页

作者： Bartsch, Alison Farimani, Amir Barati Carnegie Mellon University Department of Mechanical Engineering United States

Deformable object manipulation remains a key challenge in developing autonomous robotic systems that can be successfully deployed in real-world scenarios. In this work, we explore the the task of sculpting clay into 3D shapes. We propose the first coarse-to-fine autonomous sculpting system in which the sculpting agent first creates a coarse shape, and then iteratively refines the shape with sequences of deformation actions. We leverage large language models for sub-goal generation, and train a point cloud region-based action model to predict robot actions from the sub-goals. Additionally, our method is the first autonomous sculpting system that is a real-world text-to-3D shaping pipeline without any explicit 3D goals or sub-goals provided to the system. We demonstrate our method is able to successfully create a set of shapes solely from text-based prompting. For experimental videos, human evaluation details, and full prompts, please see our project website: https://***/***/hierarchicalsculpting. © 2025 IEEE.

关键词： ai-enabled robotics Art and Entertainment robotics Dexterous Manipulation

来源：评论

学校读者我要写书评

暂无评论

DRAGON: A Dialogue-Based Robot for Assistive Navigation With Visual Language Grounding

引用

IEEE robotics AND AUTOMATION LETTERS 2024年第4期9卷 3712-3719页

作者： Liu, Shuijing Hasan, Aamir Hong, Kaiwen Wang, Runxuan Chang, Peixin Mizrachi, Zachary Lin, Justin McPherson, D. Livingston Rogers, Wendy A. Driggs-Campbell, Katherine Univ Illinois Dept Elect & Comp Engn Champaign IL 61820 USA Univ Illinois Dept Appl Hlth Sci Champaign IL 61820 USA

Persons with visual impairments (PwVI) have difficulties understanding and navigating spaces around them. Current wayfinding technologies either focus solely on navigation or provide limited communication about the environment. Motivated by recent advances in visual-language grounding and semantic navigation, we propose DRAGON, a guiding robot powered by a dialogue system and the ability to associate the environment with natural language. By understanding the commands from the user, DRAGON is able to guide the user to the desired landmarks on the map, describe the environment, and answer questions from visual observations. Through effective utilization of dialogue, the robot can ground the user's free-form language to the environment, and give the user semantic information through spoken language. We conduct a user study with blindfolded participants in an everyday indoor environment. Our results demonstrate that DRAGON is able to communicate with the user smoothly, provide a good guiding experience, and connect users with their surrounding environment in an intuitive manner.

关键词： Human-centered robotics natural dialog for HRI ai-enabled robotics

来源：评论

学校读者我要写书评

暂无评论

RobotGPT: Robot Manipulation Learning From ChatGPT

引用

IEEE robotics AND AUTOMATION LETTERS 2024年第3期9卷 2543-2550页

作者： Jin, Yixiang Li, Dingzhe A, Yong Shi, Jun Hao, Peng Sun, Fuchun Zhang, Jianwei Fang, Bin Samsung Res China Beijing SRC B Beijing 100028 Peoples R China Tsinghua Univ Dept Comp Sci & Technol Beijing 100190 Peoples R China Univ Hamburg D-20148 Hamburg Germany Beijing Univ Posts & Telecommun Beijing 100876 Peoples R China

We present RobotGPT, an innovative decision framework for robotic manipulation that prioritizes stability and safety. The execution code generated by ChatGPT cannot guarantee the stability and safety of the system. ChatGPT may provide different answers for the same task, leading to unpredictability. This instability prevents the direct integration of ChatGPT into the robot manipulation loop. Although setting the temperature to 0 can generate more consistent outputs, it may cause ChatGPT to lose diversity and creativity. Our objective is to leverage ChatGPT's problem-solving capabilities in robot manipulation and train a reliable agent. The framework includes an effective prompt structure and a robust learning model. Additionally, we introduce a metric for measuring task difficulty to evaluate ChatGPT's performance in robot manipulation. Furthermore, we evaluate RobotGPT in both simulation and real-world environments. Compared to directly using ChatGPT to generate code, our framework significantly improves task success rates, with an average increase from 38.5% to 91.5%. Therefore, training a RobotGPT by utilizing ChatGPT as an expert is a more stable approach compared to directly using ChatGPT as a task planner.

关键词： Chatbots Robots Codes Task analysis Robot learning Shape Natural languages ai-enabled robotics learning from demonstration manipulation planning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：