Adaptive trajectory planning for scalable multiple robot control

Abstract

In this paper, we consider multiple robot display such that when users give a sequence of images as input, a large number of robots construct a sequence of formations to visualize the input images collectively. Users perceive a higher quality of responsiveness when robots respond to users` input in a more prompt and/or regular manner. A key problem in multiple robot display is trajectory planning, which determines the robot`s position and velocity along a given path while avoiding collisions. The response then consists of trajectory planning computation time and robot moving time. In the multiple robot display, there is a conflict between reducing computation time and reducing robot moving time, and it is complicated to predict such a tradeoff over various environments. We propose an adaptive framework that exploits this tradeoff by dynamically adjusting a specific parameter - how many robots are under investigation one by one in sequence for collision resolution. Our simulation results show that our framework effectively reduces the response time in a scalable manner, and it produces balanced solutions adapting to various environments. Our experiment results also show that our framework is valid with real robot settings.