基于改进LSTM-NN的安全性自动驾驶换道轨迹规划模型

为改善现有的自动驾驶换道轨迹规划模型产生的换道轨迹与真实的换道轨迹存在较大偏差的问题，提出了一种改进LSTM-NN的安全敏感性深度学习模型，该模型可以缓解当前自动驾驶轨迹规划存在的不足，输出轨迹既保证了较高的精度又提高了安全性。CarSim仿真软件模拟了本模型产生轨迹的可跟踪性，结果显示轨迹非常平滑，并且自动驾驶车辆可以高效、安全地完成换道。

A Lane-Changing Trajectory Planning Model for Automated Vehicles Based on Improved Safety-Sensitive LSTM-NN

To reduce the deviations between the planned and real lane-changing trajectories, this paper proposes an improved safety-sensitive LSTM neural network for lane-changing trajectory planning of autonomous vehicles.The trajectories generated by this model were simulated using CarSim.The results show that the produced trajectories are very smooth and the automated vehicle can complete the lane changing process efficiently and safely following the trajectory planned.The proposed deep learning model produces more accurate output trajectories ensuring higher level of safety.