Autonomous Task Allocation and Stepwise Rescheduling for Dual-Arm Robot in an Unstructured Environment

For multiple robots to perform tasks safely and efficiently in an unstructured environment, an optimal schedule is pre-planned on the basis of constraints between the tasks. However, the schedule may become inefficient or even disrupted due to deviation between the expected and actual work environment. In this study, we have developed a planning method for efficient scheduling and a management method for safe task allocation for processes involving a dual-arm robot performing a high-load task and a high-precision task simultaneously. In addition, we developed a stepwise rescheduling system with anytime A* to prevent makespan delays caused by the planning time required for online rescheduling. This was implemented in a system for anchor insertion in construction and maintenance. The results show that online re-planning eliminates the deviations between the planned and actual schedule caused by a disruptive event. The makespan of the re-planned schedule was reduced by 11% (from 776 [s] to 699 [s]) compared to when A* was used, demonstrating the effectiveness of the proposed method.