基于SAC算法的多源信息融合机械臂控制方法

Control Method of Multi-Source Information Fusion Manipulator Based on SAC Algorithm

融合多源传感器信息可实现机械臂的精准控制。提出了一种基于深度强化学习的多源信息融合控制方法,设计并搭建了可实现多源信息融合的机械臂控制实验平台,融合视觉、触觉、编码器3种模态信息,并选取学习效率较高的SAC算法实现机械臂自主精准控制。在实验平台上,开展了3个由易到难的机械臂运动控制任务,以验证所提多源信息融合机械臂控制方案的有效性。

It is necessary to integrate multi-source sensor information in the design of robotic arm to obtain various information and achieve precise control.This paper proposed a control method of muti-source information fusion manipulator based on the SAC algorithm.And this paper designs an experimental platform for manipulator control that can realize multisource information fusion.The platform can fuse the three kinds of modal information,including vision,touch and encoder,and select the SAC with high learning efficiency to realize the autonomous and accurate control of manipulator.On the experimental platform,three manipulator motion control tasks from easy to difficult are carried out to verify the effectiveness of the proposed multi-source information fusion manipulator control scheme.