Human-Machine Collaborative Path Planning Based on Eye Movement Data

Human-machine collaborative driving system can effectively improve driving safety and traffic capability. However, how to implement safety-enhanced path planning in complicated environments with potentially risk obstacles remains a challenge. In this paper, the eye tracker outputs the driver's viewpoints on the environment, so as to be used to identify the danger degree of the obstacles and make local path planning. The visual field of driver is divided into five parts, and the distribution characteristics is analyzed, which helps to determine whether an area is dangerous or not. Meantime, the bounding box is used to approximate the outline of the risk obstacle, and an expansion coefficient is set to indicate the risk degree. The expanded box forms the risk obstacle layer, which can be added to the multilayer map. For the planning modules, the global path after deformation is used as reference, and the local planning path is then obtained by multistage state space sampling. Finally, the experiment verifies that the path planning module of the human-machine collaborative driving system based on eye movement data can output safer planning results in the environment with risk obstacles, which can effectively improve driving safety.