A genetic multi-robot path planning (GMPP) algorithm is proposed to rationalize the path planning of multi-robots in multi-obstacle scenarios, especially in large and complex environments with path length constraints and path safety feasibility. On the basis of ensuring path safety, the GMPP uses the shortest path length as the objective function. Secondly, the GMPP considers the task execution problem of multi-robot path planning to avoid collisions in a shared working environment. As a result, the GMPP introduces a central planner to make collision-free driving between multiple robots through two mechanisms: collision detection and collision elimination. Comparison experiments between the GMPP and other swarm optimization algorithms are carried out. The results show that the GMPP has better overall performance in terms of anti-collision safety and efficiency for multi-robot path planning.