The traditional SLAM algorithm model is lack of the ability to describe multiple sensor information accurately in the clutter environment

and is prone to false data association. To deal with this problem

a novel SLAM algorithm based on the probability hypothesis density (PHD) filter is proposed. It models the sensor observations and environmental map as random finite sets in every time step

and constructs joint target state variable. The new algorithm estimates the robot’s poses and environmental map simultaneously through the PHD filter

and the PHD filter is realized by particle filter. To avoid the error caused by cluster algorithm

it uses a time-delay particle set outputting approach in joint target state extracting. The new algorithm model can depict the observation uncertainty

loss detecting

false alarm due to clutter and other sensor information more naturally and accurately

it also avoids the data association

the system state estimation is closer to real values. The simulation results show that the accuracy of the new algorithm in vehicle localization and mapping is improved by more than 50% compared with traditional SLAM algorithm. This algorithm provides a new solution to SLAM problem in the clutter environment.