Research On The Vehicle’s Posture Deviation Based On The Pre Vehicle Perception System

With the ability of active suspension,emergency rescue vehicles can achieve stable attitude and smooth driving under some complicated condition.These enable themselves to avoid indirect damage to the wounded and some precision rescue equipment,in addition to improving the speed of vehicles and ensuring the timeliness of rescue.Therefore,it is necessary for the vehicle to have the ability to perceive the road information in front of the vehicle in order to improve the response speed of the active suspension to the road characteristics.Based on the national key R&D project "Research on Key Technologies of Chassisand Suspension for High-mobility Emergency Rescue Vehicles(including fire-fightingvehicles)"(project number: 2016FYC0802902),this paper discusses the application and optimization of pre-vehicle perception system,which based on lidar and IMU(Inertial Measurement Unit),in estimating the body posture,ensuring the vehicle heave accuracy,reconstructing the terrain data and extracting the posture deviation that makes the car body stable.The main achievements are as follows:(1)Pre-vehicle perception system is built,which uses lidar and IMU for environment detecting,supplemented by a vehicle as a carrier unit and an on-board computer as a calculation unit.For the point cloud data acquired by lidar and the posture data measured by IMU,data preprocessing is performed to ensure data reliability and availability,which includes point cloud filtering and denoising,point cloud segmentation clustering,ground information extraction,and pose interpolation based on timestamp alignment,etc.t.(2)A pose estimation algorithm,called(35)?-ICP,is proposed which uses IMU interpolation pose data as a priori information,supplemented by line feature points and surface feature points in two frames of point cloud data for feature matching.In order to avoid unnecessary calculation constraints during optimization and ensure operation efficiency,an unconstrained optimization function with pose Lie algebra as variables is constructed,and the problem is solved using the nonlinear optimization method of LM(Levenberg–Marquart)finally,the results are weighted to obtain the final pose information.(3)Aiming at the problem of cumulative error caused by pose extraction,a global optimization method of pose graph(Pose Graph)is proposed to ensure the global consistency of the terrain data reconstructed by the optimal pose.Using VREP(Virtual Robot Experimentation Platform)robot simulation software and ROS joint simulation,the registration algorithm and global optimization algorithm are simulated and verified.Simulation results demonstrate that the terrain point cloud data reconstructed with optimized pose has no local ghosting and splicing errors,ensuring its global consistency.Under the condition of flat road,the error of detected body heave is-0.015~0.018 M,and the intersection error of trajectory is 0.01 m,which ensure the accuracy of heave and fit the expectation.(4)Gridmap is used to build the grid elevation map to ensure the efficiency of data storage.In order to avoid the complexity of pose decoupling and the high calculation dimension,the extraction technique of pose deviation based on local grid elevation map is proposed.According to the principle of three-axis all-wheel steering,the driving track of each axle wheel is estimated,and the vehicle body stability deviation is extracted according to the track information.The validity of the idea is verified by VREP simulation.(5)In this paper,pre-vehicle perception system and the method of extracting pose deviation is verified by experiments.The results show that the posture information based on feature registration and global optimization is stable and reliable,and the global map established by this has global consistency.And the heave error is maintained in the range of-0.0445~0.0591 m,which ensures the accuracy requirements of construction.Therefore the idea of extracting pose deviation based on the pre-vehicle perception system has certain practical value and significance,which realized by the following steps,the road information is reconstructed by optimal pose which calculated by point cloud registration and global optimization,and the vehicle pre-travel trajectory is estimated according to the road surface information combined with the technology of three-axis full wheel steering.