A path planning scheme for mobile robot

This paper proposes an eight-direction search scheme for mobile robot path planning, addressing dynamic obstacle avoidance in complex environments where traditional algorithms like A* and Dijkstra suffer from high complexity and inefficiency. Obstacles are modeled via neural networks as convex polygons or circles, generating collision energy distributions through boundary integration. The approach initializes a straight-line path divided into segments, using eight-directional searches with variable step lengths to enhance smoothness and efficiency. Path energy, based on squared node differences, balances avoidance and length.Simulations validate that energy weights (0.17:0.83 for path/obstacle) and variable steps cut iterations (451 to 387), remove mutations, and yield smooth, collision-free paths in static scenarios. Future extensions goal dynamic environments.