Path planning for automated robot painting

This paper describes the analysis work underlying the path-planning algorithm for a robotic painting system. The system requires no bespoke production tooling and fills an automation gap in rapid prototyping and manufacturing technology that is currently occupied by hand painting. The system creates images by exposing individual pixels of a photographic coating with a robot-mounted laser. The painting process requires no physical contact so potentially images could be developed on any shape regardless of its complexity: As objects can only be "painted" when their surface can be "hit" (i.e. exposed) by the light beam the system requires six degrees of freedom to ensure all overhanging or reentrant areas can be exposed. The accuracy of serial robots degrades with the length of the kinematic chain (in other words six axis robots cannot position themselves with the same accuracy as four axis ones). Consequently to ensure high precision in the location and orientation of the light source, the object being exposed is mounted on a rotary tilt table within the workspace of a fouraxis robot. This gives a six-degree of freedom positioning system composed of two separate kinematic chains. Although the resulting system is accurate the problems of constructing a coordinated path that allows the light beam to efficiently sweep (i.e. cover) the surface regardless of its geometry are challenging. This paper describes the difficulties and, after reviewing existing path planning algorithms, a new algorithm is introduced firstly by describing the nature of the system's configuration space and then further developing this concept as an alternative to a previously described planning algorithm. Having outlined the approach the paper presents a kinematic model for the system and compares the configuration space approach to a purely Cartesian planning approach. Copyright © 2007 by ASME.