Aiming at the limited space with complex obstacle distribution and closed boundary, an optimal path planning
algorithm based on adaptive regional grid is presented. Firstly, the environment is divided into regional grid adaptively,
and a measure is proposed to optimize the division of regional grid, which is used to reduce the dimension of search space
and defined as block degree. Afterwards, an improved particle swarm optimization(PSO) algorithm combined with random
mutation operator and directional mutation operator is proposed. Finally, the least-square curve fitting method is used to
smooth the optimal path. Simulation results show the superiority of improved PSO algorithm by comparing with nonlinearly
decreasing weight PSO(NDW-PSO) algorithm and composite PSO(C-PSO) algorithm.