Research On Human-Like Trajectory Planning And Trajectory Tracking Control For Autonomous Vehicles Based On The MUEAVI Test Track

Intelligent driving vehicles could reduce traffic accidents,improve vehicle safety,alleviate traffic congestion,reduce driving fatigue,and reduce fuel consumption to some extent,hence to bring convenience to people’s daily use of cars.In recent years,autonomous vehicles have received extensive attention from major automobile manufacturers and related research institutions.The intelligentization,connectivity and electrification of vehicles are the development direction of the automobile industry in the future,and intelligent driving vehicles have broad prospects.The key technologies to realize autonomous driving of intelligent vehicles mainly include environment perception,decision planning and control execution,and trajectory planning plays an important role in realizing the autonomous driving process.In addition to security considerations,acceptability is also an important factor for driverless vehicles promotion in the market in the future.The trajectory that reflects the driving habits of human drivers and conforms with people’s driving intuition enables a vehicle to operate smoother and more comfortable when passing through corners,which could improve the acceptability of autonomous vehicles in the market in the future.The research of this paper focuses on planning a human-like trajectory along a road based on the MUEAVI(Multi User Environment Autonomous Vehicles Innovation)track that could reflect natural driving behaviour in corners considering the sense of natural and comfortable for the occupants.Firstly,the data collected of the MUEAVI test track was processed and the coordinate system transformation was completed,and the human tested trajectories in the test track was extracted and processed.The local NED Cartesian coordinates of the test track were obtained by processing the raw data of the lane lines collected.Then the coordinate transformation from Cartesian coordinate system to Frenet coordinate system was completed,which prepared the trajectory planning and trajectory tracking research in the test track afterwards.The human driving trajectories data in the track of were collected,the driver’s single lap trajectory data and the driver’s test trajectory data on a straight road section and three corner sections are extracted after data processing.Secondly,human-like trajectories planning was completed at three corners based on the test track.A simplified mass point kinematics vehicle model was established,and the coordinate transformation to Frenet coordinate system was completed.The kinematics vehicle model with S as the independent variable was obtained to facilitate the subsequent trajectory planning in the test track,the corresponding objective function and constraint conditions were set respectively in the curve section C1,C2 and C3,constructed the optimal control problem,and then the problems were solved by the solver tool GPOPS – II,planned the human-like optimal trajectories in corners C1,C2 and C3 respectively.Finally,the trajectory coordinates in the Frenet coordinate system were transformed to the cartesian coordinates.Thirdly,Time to Line Crossing(TLC)models were derived,and the generated human-like optimal trajectories were compared with the human driving tested trajectories.The TLC calculations of lane crossing in straight road section,and straight road crossing,inside line crossing and outside line crossing in corner road section were derived.By comparing and studying the different TLCs in the three corner sections,the consistency of the human tested trajectories and the planned human-like optimal trajectories was verified.To a large extent,the human-like optimal trajectories generated reflects a similar trend with the human tested trajectories in corners,and reflects the characteristics of human drivers when people driving vehicles in corners.Then,the trajectory tracking control algorithm based on LQR was designed,and a Car Sim /Simulink co-simulation platform was established to track the human-like trajectory generated in corner C2/C3 and lane centreline trajectory.The trajectory tracking error was controlled under 0.15 m,and the control effect is acceptable.The vehicle states were compared to verify the superiority of the planned human-like optimal trajectory.Vehicles would operate smoother and more comfortable when driving along the planned human-like trajectory,which in line with people’s driving habits in daily life.Finally,the Hi L test bench based on VCU was established,including NI real-time simulator,D2 P controller,host computer,cabinet,etc.,and the trajectory tracking control test was completed to verify the effectiveness and feasibility of the trajectory tracking control algorithm designed in the real controller.