Ant Colony Optimization based Multi-Robot Planner for Combined Task Allocation and Path Finding

Abstract

Nature has inspired many solutions to the problems in computer science and recently in the field of robotics as well. Ant based algorithms have been successful in solving the NP hard problems such as traveling salesman problem. In the field of multi-robots it has been used to solve path finding and task allocation problems. In industrial warehouse applications, these problems are often combined, when for example multiple robots need to pick-up objects from one location and dropoff at the other. Multiple mobile robots need to perform these task optimally and simultaneously being on the floor without collisions. In this paper, we address this problem keeping the objective of being able to obtain collision free paths for all robots in a map, assigned for all given pick-up and drop-off tasks among themselves with an optimized minimal total distance traveled by the robots. We propose a multi-robot planner inspired from ant colony optimization to solve this combined problem. This planner finds collision free paths to all tasks to be done using a spread of ants from each robot. Ignoring the ones with collisions from other ants in their determined paths, the planner rates the tasks according to the total distance traveled. Using this rating system through multiple iterations, the planner eventually selects the best task allocations with paths for all robots among given iterations. This planner or as we call it, Ant Colony Optimization based Multi-Robot Planner for Combined Task Allocation and Path Finding Ant Colony Optimization based Multi-Robot Planner for Combined Task Allocation and Path Finding (ACTF) for pick-up and drop-off tasks is presented in this paper and has been tested against other similar planners producing promising results.