Optimal Exploration in Unknown Environments
Abstract
This paper presents an algorithm that optimally
explores an unknown environment with regions of varying
degrees of importance. The algorithm, termed
Ergodic Environmental Exploration (E³), is a finite receding horizon optimal control algorithm that minimizes control effort and the difference
between the time average behavior of the system’s trajectory
and the distribution of the gain in information. The novelty of
the E³ algorithm is the gain in information distribution used in
the exploration trajectory optimization. The gain in information
distribution uses an estimate of the information distribution and
the confidence value on that estimate. Successful experiments
have been conducted using E³ on a real mobile robot to explore
an unknown 2-dimensional area. Results of these experiments
are discussed and displayed with figures and a movie.