A Path Correction Method Based on Global and Local Matching for Robotic Autonomous Systems.

Normally, small batch parts, especially by manual assembly, are difficult to be manufactured automatically(e.g., welded, polished, sprayed) by robots because of deviation from a pre-taught path, which results from shape error and clamping error. To correct the pre-taught path, visual sensors can be used to detect it and guide the robots. However, some of the paths cannot be continuously detected due to work space limitation, and therefore, only part of them can be detected. As a result, it is difficult to directly match the incompleted detected path to pre-taught path with location errors and shape errors, which cause residual deviation after correction. This paper proposes an efficient and robust path correction method, that can be applied to robotic autonomous systems. The main contribution of this work is adopting global matching to maintain the original shape of the pre-taught path, which deals with location errors, and the local matching relaxes this constraint and addressed the shape errors. Structured light vision is used to detect the deviation of path to improve the detecting accuracy and reduce detection data size simultaneously. And a path-matching method taking account of both global and local features is developed based on iterative closest point algorithm (ICP). The proposed method consists of four steps: path scanning, global matching, local matching and data updating. Several robotic welding experiments are applied to verify the proposed method performance. [ABSTRACT FROM AUTHOR]