Method | Representative model | Advantages | Disadvantages |
---|---|---|---|
Traditional method | Artificial potential field | The planned path is smooth and safe | May fall into a local potential field, i.e., a local optimal solution |
Machine learning | Q-learning | It is a model-free method with strong adaptability and can deal with uncertain environments | Blind exploration prolongs training time, causing the agent to converge to the wrong solution |
Heuristic search | A* algorithm | High search efficiency, good stability, and can quickly respond to scene changes | Ignoring the node constraints of the moving volume, and the path planning is relatively rough |