A High Accuracy Extraction Method of Point Cloud Vectors for Guiding Robotic Arm Movement

A path planning method based on point cloud for robotic arm movement was proposed, which provides the 3D point and its vector direction for each moving position. First, an effective filtering algorithm is utilized to smooth the original point cloud. Then, the least squares method is employed to perform plane fitting on local regions of the filtered point cloud. Finally, the points with the corresponding normal vectors could be obtained for guiding the robotic arm movement. Experiments were designed to evaluate our method. Two filtering algorithms are carefully compared and chosen to simultaneously guarantee the accuracy of the gathered points and normal vectors, an evaluation metric is proposed. The point cloud of a standard sphere was acquired and fitted to calculate the sphere center. Each point and its normal vector determine one spatial line. The distance from the sphere center to each spatial line is defined as the quantitative indicator of the evaluation metric. Experimental results proved that the proposed method and evaluation metric are feasible and effective for precise motion control of robotic arms.