Background
Manufacturing and sustainment processes – such as painting, sanding, and non-destructive inspection – require automation solutions that can traverse the surface of a part in a coverage pattern with a tool. Depending on the application, these manufacturing processes can tolerate a great deal of flexibility in the way the coverage pattern is executed, while maintaining high process quality. SwRI’s experience is that typical advanced process planning imposes many artificial constraints by considering only the nominal tool path without accounting for any flexibility that the target process can accommodate. As a result, many process plans fail due to the inclusion of unreachable waypoints or segments, and those that do not fail often include costly reconfiguration moves, which negatively impact process quality.
Approach
In this project, SwRI developed a high-level process-aware planning framework capable of exploiting flexible process constraints defined by a user. These constraints include the specification of tool center point (TCP) frames and motion groups, modification strategies for tool paths, rules defining waypoint feasibility, rules for restructuring tool paths after the removal of a waypoint, and motion planner configuration. The planning framework uses these constraints to iteratively evaluate the feasibility of the tool paths, restructure the tool paths as necessary to maximize task completion, and create process motion plans for all possible execution configurations. Once the planning of process motions is complete, the framework plans motions to transition between tool paths. Finally, it synthesizes an ordered, minimum cost set of process and transition motions that accomplishes the task.
During research, development, and testing, the SwRI team created a number of commonly used, generic constraints and implemented the motion planners in the SwRI-developed Tesseract and TrajOpt software libraries. SwRI also created tools for visualizing the steps of the planning framework for validation.
Accomplishments
The planning framework was able to produce process plans for previously infeasible tool paths, such as those containing unreachable waypoints or waypoints in collision. Additionally, the framework generated more optimal motion plans with a higher rate of success, compared to the previous planning strategy, given its ability to evaluate all possible execution strategies.
After completing development of this project, the planning framework was utilized for the success of the Mobile Autonomous Coating Application for Aircraft Sustainment project, sponsored by the Advanced Robotics for Manufacturing (ARM) Institute. The target application of the project was robotic painting of aircraft components, using a spray gun with three discrete nozzle positions mounted on a 12-foot-long extension tube. Initial motion planning attempts without the process-aware planning for this application failed because the size of the components to be painted required that different nozzle positions be used in various locations on the component. Additionally, the geometry of the parts and the painting tool caused some regions of the components to be unreachable. Using the new planning framework, the robotic system was able to avoid infeasible regions of the proposed tool path and utilize the best nozzle position for each stroke of the coverage pattern. This enabled the robot to paint the maximum possible area, while minimizing the total execution time and number of nozzle re-configurations.