Potential equilibrium multi-Bernoulli filtering SLAM method based on multiple robots

The invention discloses a potential equilibrium multi-Bernoulli filtering SLAM method based on multiple robots. The method comprises the following steps of (1) establishing an observation and map feature multi-Bernoulli RFS model; (2) converting an SLAM problem into independent robot pose state estimation and map feature state estimation; (3) obtaining a robot pose state estimation predicted value at the moment k; (4) obtaining map feature state estimation at the moment k by using a potential equalization strategy; (5) trimming and combining the updated Gaussian items; (6) correcting map feature state estimation by using a Gaussian item; (7) the Gaussian item being fused into each robot observation set and substituted into the robot prior information set at the k+1 moment; (8) using the filter in the step (4) to estimate and adjust the weight of each particle map of the robot, and using particle weighted average to update the pose state estimation predicted value at the moment k; and (9) executing the step (3