Research On Dynamic Path Following Considering Collision Avoidance For Multi-intelligent Vehicles

When it comes to the area access problem of multi-intelligent vehicles(multi-IV), after the central planning system completes task allocation and path planning, each vehicle starts to follow the corresponding global path. However, the global paths generated by the path planning process are based on the known static environment, that is, only collisions between vehicles and static obstacles are considered. Once two vehicles lie close to each other, or their corresponding paths cross ahead, collision may occur between the two vehicles if the vehicles still keep the current moving state. Therefore, a dynamic path following control system, which coordinates the vehicle’s own motion in real time according to other vehicles’ information, is essential to achieve collision avoidance with the other vehicles in the dynamic environment, so as to guarantee all vehicles to fulfill the dynamic path following task in a collision-free and safe way.A dynamic path following control system considering collision avoidance among intelligent vehicles is designed in this thesis. The system consists of three modules: collision prediction module, collision avoidance module and global path following module.Based on other vehicles’ motion information, the single vehicle’s collision prediction module adopts algorithms of E-PFM and the improved VPH+, to predict possible collisions among the vehicles in a dynamic environment when moving along the global path, and then outputs two behavior patterns according to the prediction: collision avoidance and global path following.If the vehicle’s behavior pattern switches to collision avoidance, which means a possible collision will occur along the global path in the future, the collision avoidance module is triggered at once. With the proposed safe function and speed control law, the module outputs efficient collision avoidance measures in the form of real-time vehicle control parameters: the desired steering angle and the desired speed to the bottom control system to execute.If the vehicle’s behavior pattern switches back to global path following, that is, no possible collisions will occur along the global path for the time being, the global path following module is triggered. With the algorithm of Pure Pursuit, the module outputs real-time vehicle control parameters: the desired steering angle and the desired speed, and sends them to the bottom control system so as to fulfil the path following task.The simulation scenario of dynamic path following considering collision avoidance of multi-intelligent vehicles is built in V-REP, and the co-simulation with V-REP and Visual Studio 2010 verifies the proposed system’s safety and reliability.In the end, based on two BYD intelligent vehicles, which are both equipped with GPS-INS to get vehicle’s pose and velocity information, as well as GE MDS SD9 radio to communicate, the proposed system is verified to be able to be applied in real dynamic path following scenario for multi-intelligent vehicles feasibly.