An improved bicubic interpolation SLAM algorithm based on multi-sensor fusion method for rescue robot

Abstract

Aimed at the problem of difficulty in constructing an environmental map accurately with high speed and few environmental features for the rescue robot, an improved bicubic interpolation algorithm is proposed with a multi-sensor fusion method for the Hector simultaneous localisation and mapping (SLAM) in this paper. First, the robot position was estimated by the multiple information fusion based on the measurements with the extended Kalman filtering framework. Second, with the estimation results participating in the scan matching as the initial value of nonlinear optimisation, an improved bicubic interpolation algorithm is proposed instead of the bilinear interpolation algorithm for the gradient function optimisation of the laser point occupancy in the raster map. Then, the robot position calculating and the raster map updating would be done, and the problem of difficulty in accurately self-positioning and map overlapping could be resolved with environmental features lacking. Finally, the proposed algorithm is analysed and verified. The experiment results show that the proposed algorithm has good performance.