Aiming at the problem of amount of calculation of particle filtering positioning and excessive memory usage

an accelerated particle filtering method based on raster map preprocessing is proposed to enable autonomous mobile robots to quickly locate and reduce memory usage. In the map preprocessing stage

according to the set map resolution

the distance of the nearest obstacle is calculated and stored for each angle of each coordinate

and the result of the lookup table is obtained; in the particle filtering stage

the graphics processing unit（ GPU） is used to maintain the particles in parallel. The distance information of obstacles around particles can be searched from the improved lookup table

and then used to calculate the weight of particles. The experimental results show that the method occupies less memory and the positioning speed is also significantly improved.