节点文献
基于递归神经网络的多机器人智能协同控制
Intelligent Cooperative Control of Multiple Manipulators Based on Recurrent Neural Network
【摘要】 多机器人协同系统具有负载能力强、工作空间广、灵活性好等优点,是机器人领域的研究热点。由于系统非线性、关节约束等问题,给其高性能协同控制带来了巨大挑战。研究关节约束下多机器人智能协同运动控制问题,设计同步策略并将多机协同控制问题建模为一个二次型优化问题,基于速度逃逸法将机器人关节角与关节角速度限幅归并描述到速度层,基于递归神经网络(RNN)设计了一个稳定性可证明的实时求解器。MATLAB与V-REP联合仿真平台下以多LBR iiwa机器人系统为对象的仿真实验表明,基于所提协同控制策略所有机器人都精确跟踪到了期望的三环路径轨迹,跟踪误差可达10-4量级,轨迹跟踪过程中关节角、关节角速度皆在约束限幅内。
【Abstract】 Multi-robot cooperative system has the advantages of high-load capacity, wide working space and good flexibility, and therefore has become the hotspot in the field of robotic control. Due to the nonlinearity and joint constraints, it is a great challenge to achieve high-performance cooperative control of multi-robot systems. Intelligent cooperative control of multi-robot systems under joint constraints was considered. The basic problem was formulated as a quadratic programming problem. Based on the velocity escape method, both the joint angle and joint velocity limits were uniformly built in the velocity level and described as an inequality constraint. A stability provable online solver based on recurrent neural network was designed. Numerical experiments of the multi-LBR iiwa robot system on MATLAB-VREP co-simulation platform show that all robots accurately track the desired three-ring path trajectory based on the proposed control strategy, with the tracking error being10-4 level, and both the joint angle and joint speed values satisfy the setting joint constraints.
【Key words】 multiple robot system; cooperative control; recurrent neural network;
- 【文献出处】 机电工程技术 ,Mechanical & Electrical Engineering Technology , 编辑部邮箱 ,2020年05期
- 【分类号】TP242;TP183
- 【被引频次】6
- 【下载频次】495