Kinect sensor-based trajectory planning method of collision avoidance for industrial manipulator with an dexterous hand

Abstract

This paper presents Kinect sensor-based trajectory planning method of collision avoidance for an industrial manipulator with multi-fingered dexterous hand, which is used for grasping objects by using a data glove in dynamically changing environments. Each finger’s joint angles of the dexterous hand are synchronously collected by the data glove and forwarded to the trajectory planning control system that navigates the dexterous hand to the grasping poses. In order to effectively calculate the solutions to the desired configuration of the manipulator with the dexterous hand, the LazyPRM algorithm is also applied to maintain a road map for answering multiple planning queries. To maintain the dynamically changing environments, the Kinect sensor is used to detect and track the moving objects that appear in the workspace and to update the environment representation to avoid the collisions. The distances between the obstacles and the camera are obtained by the depth camera of the Kinect sensor. The experimental results have verified the effectiveness of the methods.