An incremental SLAM algorithm with backtracking revisable data association for mobile robots

TLDR: The necessity of revisable data association for SLAM is analyzed and graph search of AI is used to model and solve the revising data association problem.

Abstract: This paper illustrates the reason why revisable data association is needed for the simultaneous localization and mapping (SLAM) of mobile robots, and an incremental SLAM algorithm with backtrack searching data association is presented. Our approach uses a tree model called correspondence tree (CT) to represent the solution space of the data association problem. CT is layered according to time steps and every node in it is a data association hypothesis for the measurements gotten at-a-time. A best-first with limit backtracking search strategy is designed to find the optimal path in CT. A state estimation method based on the least-squares problem is developed. This method can compute the cost of nodes in CT and update state estimation incrementally, so direct feedback is introduced from the state estimation process to the data association model. With the interaction between data association and state estimation, and combining with tree pruning techniques, our approach can get accurate data association and state estimation for online SLAM applications. The contribution of this paper is that we have analyzed the necessity of revisable data association for SLAM and we use graph search of AI to model and solve the revising data association problem.