Data-driven Spatial Classification using Multi-Arm Bandits for Monitoring with Robot Teams

This video summarizes our recent work on the spatial classification problem for monitoring using data collected by a coordinated team of mobile robots. Our data-driven strategy uses a combination of multi-armed bandit framework and an optimization-based, multi-agent motion planner to solve the classification problem quickly while respecting the physical constraints on the robots. This video also shows the hardware experiments using drones and ground robots to validate the approach. This work was done by Xiaoshan Lin (UMN), Siddharth Nayak (MIT), Stefano Di Cairano (MERL), and Abraham P. Vinod (MERL). Notes: 1. Paper link to be added after arxiv paper is uploaded.