决策树算法在机械臂碰撞规避方法中的应用

Application of Decision Tree Algorithm in Collision Avoidance Method of Manipulator

针对机械臂在复杂环境作业时,由于碰撞规避方法的碰撞检测效果差、障碍物规避效果不佳,导致机械臂频繁发生碰撞的问题,提出一种应用决策树算法的机械臂碰撞规避方法。通过分析机械臂运动学结构,获取机械臂运动学方程,结合二次投影法将机械臂碰撞过程,简化为三维视角下包围盒相交情况的判断过程,检测机械臂碰撞情况,通过浓度函数采集三维空间中包围盒预计位移信息,建立基于预计位移信息的决策树模型,获取机械臂碰撞规避行为最优决策信息,实现机械臂规避碰撞。实验结果表明,应用决策树算法的机械臂碰撞规避方法仅需0.29s有效检测出碰撞,在复杂环境下,规划的碰撞规避路线中机械臂与障碍物未发生碰撞,因此,该方法有效提高了机械臂碰撞规避的效果。

Aiming at the problem that the robot arm frequently collides due to the poor collision detection effect and obstacle avoid-ance effect of the collision avoidance method when the manipulator operates in a complex environment,a collision avoidance method of the manipulator using the decision tree algorithm is proposed.The kinematics equation of the manipulator was ob-tained by analyzing the kinematics structure of the manipulator.The collision process of the manipulator was simplified into the judgment process of the intersection of the bounding box in the three-dimensional perspective combined with the quadratic projec-tion method,the collision of the manipulator was detected,the predicted displacement information of the bounding box in the three-dimensional space was collected through the concentration function,and the decision tree model based on the predicted dis-placement information was established to obtain the optimal decision information of collision avoidance behavior of manipulator,the isolation between the manipulator and obstacles was realized.The optimal decision information of manipulator collision avoid-ance behavior was obtained to realize manipulator collision avoidance.The experimental results show that the robotic arm colli-sion avoidance method using decision tree algorithm only needs 0.29s to effectively detect collisions.In complex environments,there is no collision between the robotic arm and the obstacle in the planned collision avoidance route.Therefore,this method can effectively improve the effect of the robotic arm collision avoidance.