Efficient particle filter algorithm for ultrasonic sensor based 2D range-only SLAM application

Simultaneously localization and mapping (SLAM) is a fundamental topic in robotics and mobile computing communities. Owning to low cost and relatively accurate measurement, ultrasonic sensors are widely used in range only SLAM applications. But these applications are challenging to accurately obtain an initial position given to the SLAM algorithm, and robustly localize the targets in a highly dynamic environment. The traditional approaches to solve these problems are Extended Kalman Filter (EKF) and Particle Filter (PF) algorithms. However, EKF approach suffers from the data association problem and map consistency, PF approach is limited by the highly computational complexity. This paper addresses an improved particle filter algorithm to solve the ultrasonic sensor based 2D range-only simultaneously localization and mapping (SLAM) problem with relatively good accuracy and robustness. A technique called Map Adjustment is proposed to increase the accuracy and efficiency of the algorithm. Using Map Adjustment, the proposed particle filter algorithm can either achieve improved localisation accuracy, or maintain the same accuracy but lower computational complexity. The feasibility and robustness of this algorithm is shown by experiments. The results demonstrate that the proposed algorithm can provide relatively good accuracy and robustness for ultrasonic sensor based 2D range SLAM applications.