A Novel Stop Criterion to Support Efficient Multi-Robot Mapping

Abstract

The exploration of an unknown environment is a classical and essential challenge in mobile robotics. Typically, the overall exploration time is the most commonly used quality indicator. Choosing a stop exploration criterion represents a critical issue, and the simple rule of exploring until reaching a given coverage percentage is useful only on simulations, but futile in practice. In this work we propose a stop criterion based on the expected information gain of the exploring fleet; we show that every component of the fleet can autonomously check whether there are reachable cells from where new information can be gathered, and this fact can be used to take a decision. We measure the exploring area error introduced by the stop decision, concluding that the proposal is feasible, while improvable. Certainly, the stop decisions are taken autonomously, and we understand that a cooperative approach may further improve the proposal. The experiments that support this work are built using ROS over the MORSE physical simulator.