Information Efficient 3D Visual SLAM in Unstructured Domains

This paper presents a strategy for increasing the efficiency of simultaneous localisation and mapping (SLAM) in unknown and unstructured environments using a vision-based sensory package. Traditional feature-based SLAM, using either the extended Kalman filter (EKF) or its dual, the extended information filter (EIF), leads to heavy computational costs while the environment expands and the number of features increases. In this paper we propose an algorithm to reduce computational cost for real-time systems by giving robots the 'intelligence' to select, out of the steadily collected data, the maximally informative observations to be used in the estimation process. We show that, although the actual evaluation of information gain for each frame introduces an additional computational cost, the overall efficiency is significantly increased by keeping the matrix compact. The noticeable advantage of this strategy is that the continuously gathered data is not heuristically segmented prior to be input to the filter. Quite the opposite, the scheme lends itself to be statistically optimal.