Research On 2.5D Real-time Reconstruction Technology Of Road Conditions In Front Of Vehicle Based On Kalman Estimation

The special working nature and application of emergency rescue vehicles determine that they still need to maintain the ride comfort when facing complex road conditions,and an effective way to improve the ride comfort is to adjust the active suspension through the preview control based on look-ahead.At present,researches on preview control based on look-ahead are mostly focused on control algorithms,while researches on methods of acquiring input information are few.To this end,this paper reconstructs the road conditions based on the 2.5D reconstruction technology and data fusion technology,and extracts the elevation information on the predicted trajectory of the vehicle as input information for the preview control.This paper combines the national key research and development plan project "Research and Application Demonstration of Key Technologies for High-Mobility Multifunctional Emergency Rescue Vehicles"(No.2016YFC0802900)to study the 2.5D reconstruction method of road conditions and the extraction method of road conditions elevation information.The main research work is as follows:(1)The hardware construction and software design of the road conditions reconstruction system are completed.The sensors are selected according to the needs and installed in the appropriate position.Then the software design of the system is realized by programming based on the Win10 system.(2)In order to improve the stability of the system and the accuracy of the vehicle pose,a vehicle slope motion model is established,and the vehicle pose estimation is realized based on this model.Firstly,a vehicle slope motion model is established,and the influence of road slope on vehicle position estimation is analyzed based on this model.Then the EKF method is used to realize the vehicle pose estimation based on the vehicle slope motion model.Aiming at the outliers in GPS information,a method based on Mahalanobis distance to reduce the influence of GPS outliers is proposed.(3)Aiming at the problem that traditional methods cannot balance the real-time capability and accuracy of road conditions acquisition,a method of road conditions reconstruction based on 2.5D reconstruction technology and data fusion technology is proposed.Firstly,the system measurement error is analyzed and the error transmission model is established.Then the local elevation map of road conditions is constructed and updated based on the pose data and point cloud data.To improve the accuracy,the Kalman filter method is used to estimate the elevation value.Finally,the data is stored based on the Grid Map library and visualized based on the OpenGL library.(4)Considering that the input information of the preview control is elevation information of the road conditions in front of the vehicle,a method for extracting elevation information of road conditions is proposed.According to the actual situation,this paper approximates the part of the wheel that touches the ground as an ellipse,and uses the weighted average of all the grid elevation values in the ellipse range as the elevation value of this position.When selecting weights,the distance between each grid in the ellipse and the center of the ellipse is taken as the weight,that is,the larger the distance is,the smaller the weight is.(5)The validity and practicability of the method are verified by experiments.Based on the built hardware platform,the experiments of vehicle pose estimation,road conditions reconstruction and elevation values extraction are completed.The results show that the method is effective and can guarantee the accuracy of road conditions reconstruction under the condition of meeting real-time requirements.