Research On The Positioning And Control Methods For The Intelligent Following Of Unmanned Ground Vehicles

Unmanned ground vehicles(UGVs)are considered to have great potentials to signif-icantly improve the transportation efficiency,reduce traffic accidents and fuel consump-tion.Intelligent following is one of the probable application forms of UGVs in the future.The works of this dissertation are expanded around the intelligent following problem of UGVs.The relationships among the intelligent following,master-slave remote control,fully autonomous and semi-autonomous have been analyzed.The features of the intel-ligent following under different running modes are discussed and compared.Position-ing and control problems of the intelligent following have been refined.The positioning method of the leader and the follower,the longitudinal and the lateral motion control method for the intelligent following are investigated in detail.The main contributions and innovations are as follows:(1)A new position method based on the motion trajectory matching is proposed.Ac-cording to the characteristics that the reckoning trajectory based on the gyro and odometry is relatively accurate in a short distance,the shape of the reckoning trajectory is employed to restrict the results of directly positioning methods like GNSS.Accurate,stable and s-mooth positioning results can be derived continuously by the proposed method.(2)An acceleration planning method combining the trapezoidal speed curve and the NDP supplementary structure is proposed.Based on this,the longitudinal motion controller for the intelligent following of UGVs is designed.The acceleration planning method is composed of a trapezoidal speed curve based acceleration planner and an NDP supplementary structure parallelly.The trapezoidal speed curve based planner can out-put the main component of the desired acceleration.In the supplementary structure,the learned NDP supplementary acceleration planning policy can further improve the per-formance of the original trapezoidal method in the respects of settling time,steady-state accuracy,etc.(3)A path planning framework based on virtual lanes,and a path generation method considering endpoint curvature constraint and a path evaluation method based on the shortest passage time are proposed.Based on this,the lateral motion controller for the intelligent following of UGVs is designed.In the virtual lanes based framework,path planning is divided into the reference path planning phase and the maneuver path plan-ning phase.With this two-phase scheme,the motion stability of UGVs can be enhanced by limiting the change frequency and amplitude of the reference path and the entry point of the maneuver path.The path evaluation method based on the shortest passage time overcomes the problem that the physical meaning of each index is inconsistent and the weighted coefficient is difficult to design in the general evaluation method.