Intelligent Vehicle Planning And Control Strategy Based On The Prediction Of Longitudinal Ramp Traversability

In order to solve the problem that the intelligent unmanned vehicle can't judge whether to pass in the unknown ramp scene,this paper proposes a planning control strategy based on the prediction of the longitudinal ramp traversability.By collecting the current ramp information,analyzing the passing conditions,and representing the traversability in the form of probability,the judgment can be made before climbing.When it is judged that can be passed,the longitudinal uphill speed shall be planned to control the vehicle to go up the ramp safely and smoothly;when it is judged that cannot be passed,the front ramp shall be regarded as an obstacle,and the local obstacle avoidance trajectory shall be planned in real time to control the vehicle to bypass the ramp to reach the destination.The ramp driving strategy proposed in this paper provides a judgment method and corresponding scheme for the intelligent unmanned vehicle driving in the ramp environment,which significantly improves the work efficiency and greatly reduces the security risks,and has important research significance and broad application prospects.This paper mainly studies the traversability and driving strategy of the longitudinal ramp.The confidence factor is designed for mass estimation,the interactive multi model method is used for slope estimation,and the mass and slope information are used for vehicle intelligent control.The probability of traversability is calculated based on the normal distribution,and the threshold value is set for judgment.When it is possible to pass,the longitudinal speed is planned,and the PID controller is used to control the vehicle to climb the slope safely;when it is impossible to pass,the local obstacle avoidance trajectory is planned by quintic polynomials,and the MPC controller is designed to follow the desired trajectory.Finally,the whole system is tested and verified by Matlab/Simulink-CarSim joint simulation platform and real vehicle verification platform.The main research contents are as follows:(1)Vehicle mass and road slope estimationFirstly,the relevant parameters and distribution range suitable for mass estimation are determined,the confidence factor of mass estimation is synthesized by fuzzy reasoning rules,and the threshold value is set.When the confidence factor is greater than the threshold,the least square method with forgetting factor is used to estimate the mass.Secondly,standard Kalman filter based on vehicle kinematics and unscented Kalman filter based on vehicle dynamics are used to estimate the slope respectively.Because of the obvious advantages and disadvantages of the two estimation methods,the interactive multi model estimation method is used to fuse the two estimation methods to get the final slope estimation.(2)Prediction of longitudinal ramp traversabilityBased on the analysis of the three dynamic conditions and the normal distribution calculation characteristics,a prediction method for the ramp traversability based on the normal distribution is proposed.By analyzing the point cloud model of the measured ramp,it is proved that the slope angle distribution approximately obeys the normal distribution.Finally,in the environment of CarSim,a variety of ramp scenes which obey the normal distribution are built to verify the effectiveness of the traversability prediction method.(3)Longitudinal driving strategy and local obstacle avoidance strategyDifferent driving strategies are designed based on the prediction results of the ramp traversability.When it is possible to pass,a longitudinal speed planning method based on interval partition is designed,and the PID controller is used to control the driving force to realize the longitudinal speed following;when it is impossible to pass,this paper proposes a quintic polynomial trajectory planning method based on sampling,and selects the optimal trajectory by designing penalty function and relevant constraints.Then,a two freedom dynamic model of vehicle is established,and a trajectory following controller based on model predictive control is designed.The optimal function and dynamic constraints are set to control the vehicle to track the expected obstacle avoidance trajectory.(4)Test verificationBuild the MATLAB / Simulink-CarSim joint simulation test and real vehicle verification test platform,design different working conditions,and verify the real-time and effectiveness of the mass slope joint estimation algorithm.Based on the environment of MATLAB / Simulink-CarSim,design the different sizes and distributions ramp scenes.The judgment method of the ramp traversability,the corresponding longitudinal driving strategy and the local obstacle avoidance strategy are completely tested and verified.