检索结果-内蒙古大学图书馆

A Variational Bayesian Gaussian Mixture-Nonnegative Matrix Factorization model to Extract Movement Primitives for Robust control

引用

IEEE TRANSACTIONS ON HUMAN-MACHINE SYSTEMS 2022年第6期52卷 1258-1270页

作者： Xie, Hongbo Mengersen, Kerrie Di, Changan Zhang, Yongjian Lipman, Justin Van Huffel, Sabine Yunnan Univ Sch Informat Sci & Engn Kunming 650500 Yunnan Peoples R China Queensland Univ Technol Ctr Data Sci Brisbane Qld 4001 Australia Nanjing Univ Sci & Technol Sch Mech Engn Nanjing 210094 Peoples R China Univ Technol Sydney Fac Engn & IT Ultimo NSW 2007 Australia Katholieke Univ Leuven Dept Elect Engn ESAT Stadius Div B-3001 Leuven Belgium

Nonnegative matrix factorization (NMF) is a powerful tool for parameter estimation applied in numerous robotics applications, such as path planning, motion trajectory prediction, and motion intention detection. In particular, NMF has been successfully used to extract simplified and organized movement primitives from myoelectric signal (MES) for robust control of multi-degree of freedom humanoid robots. However, MES is typically contaminated by complex noise sources. The system performance often degrades due to the simplified Gaussian assumption of the noise distribution in existing NMF methods. Furthermore, most existing NMF models are unable to automatically determine the rank of the latent matrices. To address these issues, this article presents a hybrid variational Bayesian Gaussian mixture and NMF (GMNMF) model with a finite Gaussian mixture model adopted to fit the mixed noise density function of MES. In addition, the automatic relevant determination criterion is applied to automatically infer the number of movement primitives. The coordinate descent update rules for the proposed model are formulated by mean-field variational Bayesian inference. We assess the model performance on five synthetic noise distribution functions and an experimental MES dataset to perform six wrist movements. The results demonstrate that GMNMF yields low error and high robustness in extracting the movement primitives over four competitive methods for robust cybernetic control.

关键词： Manganese Bayes methods Gaussian distribution Muscles Data models Robustness Man-machine systems Bayesian inference Gaussian mixture model (GMM) humanoid robots model learning for control movement primitive nonnegative matrix factorization (NMF)

来源：评论

学校读者我要写书评

暂无评论

learning Non-Parametric models in Real Time via Online Generalized Product of Experts

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 9326-9333页

作者： Watson, Connor Morimoto, Tania K. Univ Calif San Diego Dept Mech & Aerosp Engn La Jolla CA 92093 USA Univ Calif San Diego Dept Surg La Jolla CA 92093 USA

In this work, we address the problem of online learning, where models must be continually updated from an incoming stream of data, while retaining past information. We develop an approach that is nonparametric, models uncertainty, and requires minimal hand-tuning. Our proposed algorithm, which we term online generalized product of experts (OGPoE), extends the powerful generalized product of experts (GPoE) framework to the online setting by leveraging methods for sparse, variational Gaussian process approximations, as well as nonparametric clustering. We devise a 1-D example learning problem to illustrate how our method works, and we verify that we achieve competitive results with other popular modeling approaches on a benchmark learning problem. Finally, we demonstrate how our algorithm can produce high accuracy predictions on a physical system, by learning the kinematics for a concentric tube robot, even when the robot is subject to changing, unknown loads.

关键词： model learning for control modeling control and learning for soft robots machine learning for robot control

来源：评论

学校读者我要写书评

暂无评论

Robotic Embodiment of Human-Like Motor Skills via Reinforcement learning

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第2期7卷 3711-3717页

作者： Guzman, Luis Morellas, Vassilios Papanikolopoulos, Nikolaos Univ Minnesota Dept Comp Sci & Engn Minneapolis MN 55455 USA Univ Minnesota Dept Elect & Comp Engn Minneapolis MN 55455 USA

Current methods require robots to be reprogrammed for every new task, consuming many engineering resources. This work focuses on integrating real and simulated environments for our proposed "Internet of Skills," which enables robots to learn advanced skills from a small set of expert demonstrations. By expanding on recent work in the areas of learning from Demonstrations (LfD) and Reinforcement learning (RL), we can train robot control policies that can not only effectively complete a given task but also do so with greater performance than the expert demonstrations used to train the policy. In this work, we create simulated environments to train RL algorithms for the task of inverse kinematics and obstacle avoidance. Many state-of-the-art RL algorithms are compared, and we provide a detailed analysis of the state space and parameters chosen. Lastly, we utilize a Vicon motion tracking system and train the robot agent to follow trajectories given by a human operator. Our results show that reinforcement learning algorithms such as proximal policy optimization can develop control policies that are capable of complex control tasks that integrate with the real world, an important first step towards developing a system that can autonomously learn new skills from human demonstrations.

关键词： Robots Task analysis Robot kinematics Reinforcement learning Training Collision avoidance Virtual environments Telerobotics and teleoperation reinforcement learning transfer learning model learning for control collision avoidance

来源：评论

学校读者我要写书评

暂无评论

The RATTLE Motion Planning Algorithm for Robust Online Parametric model Improvement With On-Orbit Validation

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 10946-10953页

作者： Albee, Keenan Ekal, Monica Coltin, Brian Ventura, Rodrigo Linares, Richard Miller, David W. MIT Dept Aeronaut & Astronaut Cambridge MA 02142 USA Inst Super Tecn Inst Syst & Robot P-1049001 Lisbon Portugal NASA SGT Inc Ames Res Ctr Moffett Field CA 94035 USA

Certain forms of uncertainty that robotic systems encounter can be explicitly learned within the context of a known model, like parametric model uncertainties such as mass and moments of inertia. Quantifying such parametric uncertainty is important for more accurate prediction of the system behavior, leading to safe and precise task execution. In tandem, providing a form of robustness guarantee against prevailing uncertainty levels like environmental disturbances and current model knowledge is also desirable. To that end, the authors' previously proposed RATTLE algorithm, a framework for online information-aware motion planning, is outlined and extended to enhance its applicability to real robotic systems. RATTLE provides a clear tradeoff between information-seeking motion and traditional goal-achieving motion and features online-adjustable models. Additionally, online-adjustable low level control robustness guarantees and a new method for automatic adjustment of information content down to a specified estimation precision is proposed. Results of extensive experimentation in microgravity using the Astrobee robots aboard the International Space Station and practical implementation details are presented, demonstrating RATTLE's capabilities for real-time, robust, online-adjustable, and model information-seeking motion planning capabilities under parametric uncertainty.

关键词： Constrained motion planning information-aware planning model learning for control model predictive control space robotics and automation

来源：评论

学校读者我要写书评

暂无评论

Physics-Inspired Temporal learning of Quadrotor Dynamics for Accurate model Predictive Trajectory Tracking

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 10256-10263页

作者： Saviolo, Alessandro Li, Guanrui Loianno, Giuseppe NYU Tandon Sch Engn Brooklyn NY 11201 USA

Accurately modeling quadrotor's system dynamics is critical for guaranteeing agile, safe, and stable navigation. The model needs to capture the system behavior in multiple flight regimes and operating conditions, including those producing highly nonlinear effects such as aerodynamic forces and torques, rotor interactions, or possible system configuration modifications. Classical approaches rely on handcrafted models and struggle to generalize and scale to capture these effects. In this letter, we present a novel Physics-Inspired Temporal Convolutional Network (PI-TCN) approach to learning quadrotor's system dynamics purely from robot experience. Our approach combines the expressive power of sparse temporal convolutions and dense feed-forward connections to make accurate system predictions. In addition, physics constraints are embedded in the training process to facilitate the network's generalization capabilities to data outside the training distribution. Finally, we design a model predictive control approach that incorporates the learned dynamics for accurate closed-loop trajectory tracking fully exploiting the learned model predictions in a receding horizon fashion. Experimental results demonstrate that our approach accurately extracts the structure of the quadrotor's dynamics from data, capturing effects that would remain hidden to classical approaches. To the best of our knowledge, this is the first time physics-inspired deep learning is successfully applied to temporal convolutional networks and to the system identification task, while concurrently enabling predictive control.

关键词： Aerial systems model learning for control optimization and optimal control robot learning

来源：评论

学校读者我要写书评

暂无评论

Smooth model Predictive Path Integral control Without Smoothing

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 10406-10413页

作者： Kim, Taekyung Park, Gyuhyun Kwak, Kiho Bae, Jihwan Lee, Wonsuk Agcy Def Dev Ground Technol Res Inst Daejeon 34186 South Korea Seoul Natl Univ Dept Mech Engn Seoul 08826 South Korea

We present a sampling-based control approach that can generate smooth actions for general nonlinear systems without external smoothing algorithms. model Predictive Path Integral (MPPI) control has been utilized in numerous robotic applications due to its appealing characteristics to solve non-convex optimization problems. However, the stochastic nature of sampling-based methods can cause significant chattering in the resulting commands. Chattering becomes more prominent in cases where the environment changes rapidly, possibly even causing the MPPI to diverge. To address this issue, we propose a method that seamlessly combines MPPI with an input-lifting strategy. In addition, we introduce a new action cost to smooth control sequence during trajectory rollouts while preserving the information theoretic interpretation of MPPI, which was derived from non-affine dynamics. We validate our method in two nonlinear control tasks with neural network dynamics: a pendulum swing-up task and a challenging autonomous driving task. The experimental results demonstrate that our method outperforms the MPPI baselines with additionally applied smoothing algorithms.

关键词： Optimization and optimal control planning under uncertainty model learning for control autonomous vehicle navigation field robots

来源：评论

学校读者我要写书评

暂无评论

Dynamic modeling of Bucket-Soil Interactions Using Koopman-DFL Lifting Linearization for model Predictive Contouring control of Autonomous Excavators

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第1期7卷 151-158页

作者： Sotiropoulos, Filippos E. Asada, H. Harry MIT Dept Mech Engn Cambridge MA 02139 USA

A lifting-linearization method based on the Koopman operator and Dual Faceted Linearization is applied to the control of a robotic excavator. In excavation, a bucket interacts with the surrounding soil in a highly nonlinear and complex manner. Here, we propose to represent the nonlinear bucket-soil dynamics with a set of linear state equations in a higher-dimensional space. The space of independent state variables is augmented by adding variables associated with nonlinear elements involved in the bucket-soil dynamics. These include nonlinear resistive forces and moment acting on the bucket from the soil, and the effective inertia of the bucket that varies as the soil is captured into the bucket. Variables associated with these nonlinear resistive and inertia elements are treated as additional state variables, and their time evolution is represented as another set of linear differential equations. The lifted linear dynamic model is then applied to model Predictive Contouring control, where a cost functional is minimized as a convex optimization problem thanks to the linear dynamics in the lifted space. The lifted linear model is tuned based on a data-driven method by using a soil dynamics simulator. Simulation experiments on homogeneous soil verify the effectiveness of the proposed lifting linearization compared to its counterpart.

关键词： Autonomous excavation construction and mining robots lifting linearization koopman operator model predictive contouring control model learning for control

来源：评论

学校读者我要写书评

暂无评论

TOAST: Trajectory Optimization and Simultaneous Tracking Using Shared Neural Network Dynamics

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 9747-9754页

作者： Kim, Taekyung Lee, Hojin Hong, Seongil Lee, Wonsuk Agcy Def Dev Ground Technol Res Inst Daejeon 34186 South Korea

Neural networks have been increasingly employed in model Predictive controller (MPC) to control nonlinear dynamic systems. However, MPC still poses a problem that an achievable update rate is insufficient to cope with model uncertainty and external disturbances. In this letter, we present a novel control scheme that can design an optimal tracking controller using the neural network dynamics of the MPC, making it possible to be applied as a plug-and-play extension for any existing model-based feedforward controller. We also describe how our method handles a neural network containing history information, which does not follow a general form of dynamics. The proposed method is evaluated by its performance in classical control benchmarks with external disturbances. We also extend our control framework to be applied in an aggressive autonomous driving task with unknown friction. In all experiments, our method outperformed the compared methods by a large margin. Our controller also showed low control chattering levels, demonstrating that our feedback controller does not interfere with the optimal command of MPC.

关键词： Feedforward systems Heuristic algorithms Adaptive control Trajectory optimization Vehicle dynamics Task analysis Predictive models Autonomous vehicle navigation field robots model learning for control optimization and optimal control robust adaptive control

来源：评论

学校读者我要写书评

暂无评论

Data-Driven modelling and control for Robot Needle Insertion in Deep Anterior Lamellar Keratoplasty

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第2期7卷 1526-1533页

作者： Edwards, William Tang, Gao Tian, Yuan Draelos, Mark Izatt, Joseph Kuo, Anthony Hauser, Kris Univ Illinois Dept Comp Sci Urbana IL 61801 USA Duke Univ Dept Biomed Engn Durham NC 27708 USA Duke Univ Dept Ophthalmol Durham NC 27710 USA

Deep anterior lamellar keratoplasty (DALK) is a technique for cornea transplantation which is associated with reduced patient morbidity. DALK has been explored as a potential application of robot microsurgery because the small scales, fine control requirements, and difficulty of visualization make it very challenging for human surgeons to perform. We address the problem of modelling the small scale interactions between the surgical tool and the cornea tissue to improve the accuracy of needle insertion, since accurate placement within 5% of target depth has been associated with more reliable clinical outcomes. We develop a data-driven autoregressive dynamic model of the tool-tissue interaction and a model predictive controller to guide robot needle insertion. In an ex vivo model, our controller significantly improves the accuracy of needle positioning by more than 40% compared to prior methods.

关键词： Medical robots and systems surgical robotics: planning model learning for control

来源：评论

学校读者我要写书评

暂无评论

Randomized-to-Canonical model Predictive control for Real-World Visual Robotic Manipulation

引用

IEEE ROBOTICS AND AUTOMATION LETTERS 2022年第4期7卷 8964-8971页

作者： Yamanokuchi, Tomoya Kwon, Yuhwan Tsurumine, Yoshihisa Uchibe, Eiji Morimoto, Jun Matsubara, Takamitsu Nara Inst Sci & Technol NAIST Grad Sch Informat Sci Nara 6300192 Japan Adv Telecommun Res Inst Int ATR Kyoto 6190288 Japan Kyoto Univ Dept Syst Sci Grad Sch Informat Kyoto 6068501 Japan

Many works have recently explored Sim-to-real transferable visual model predictive control (MPC). However, such works are limited to one-shot transfer, where real-world data must be collected once to perform the sim-to-real transfer, which remains a significant human effort in transferring the models learned in simulations to new domains in the real world. To alleviate this problem, we first propose a novel model-learning framework called Kalman Randomized-to-Canonical model (KRC-model). This framework is capable of extracting task-relevant intrinsic features and their dynamics from randomized images. We then propose Kalman Randomized-to-Canonical model Predictive control (KRC-MPC) as a zero-shot sim-to-real transferable visual MPC using KRC-model. The effectiveness of our method is evaluated through a valve rotation task by a robot hand in both simulation and the real world, and a block mating task in simulation. The experimental results show that KRC-MPC can he applied to various real domains and tasks in a zero-shot manner.

关键词： Deep learning for visual perception machine learning for robot control model learning for control

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：