To address the issues of unreasonable collision avoidance path planning
algorithms and inadequate safety in high-speed scenarios, a trajectory
prediction-based collision avoidance path planning algorithm has been proposed.
First, a trajectory prediction model is constructed using the long–short-term
memory (LSTM) network, and the trajectory prediction model is trained and tested
with the HighD dataset. Second, the future trajectory of the obstacle car is
predicted, the future trajectory information of the two cars is combined to
generate the lane-changing decision, and the three-times B-spline curves are
used to generate the collision avoidance path clusters. The optimal collision
avoidance paths are generated based on the multi-objective optimization
function. Finally, build a MATLAB/CarSim simulation platform to verify the
reasonableness and safety of the planned paths by taking the three scenarios of
the continuous overtaking, preceding car pulling out, and the neighboring car
cutting in as examples. The results show that the proposed trajectory planning
algorithm is more robust, reasonable than the path planning algorithm based on
the safety distance model, effectively improving the safety of the vehicle
during high-speed driving.
