Extraction of Preview Elevation Information Based on Terrain Mapping and Trajectory Prediction in Real-Time

Lili Wang,Shuang Liu,Tao Ni,Dingxuan Zhao

doi:10.1109/access.2020.2984034

Abstract

Preview elevation values are used as inputs to suspensions to improve a vehicle’s ride. In a previous study, we obtained preview terrain elevation information in post-processing. In this paper, we contributed a method to reconstruct the terrain elevation map in real-time to extract preview elevation information according to the predicted trajectory. We use interdisciplinary knowledge to solve an engineering problem in the field of mechanical engineering, and the innovation of this manuscript belongs to Engineering Application Innovation. First, in the process of data collection, the terrain elevation map is updated with LIDAR measurements. A Kalman filter is used to update the elevation map. Second, as the vehicle moves, the uncertainty in the vehicle’s motion is estimated, and the error variance that propagates to the terrain elevation map is computed. Third, the vehicle trajectory is predicted based on a vehicle kinematics model. According to the predicted trajectory, the elevation information is extracted from the terrain elevation map in advance. The proposed method has been extensively evaluated with the simulations in Gazebo and outdoor environments, and the results demonstrate that the algorithm is effective, with a maximum error of 2.58 cm for the estimated elevation value.

Highlights

While on route to rescue sites in remote areas with rough terrain, the chassis performance of emergency rescue vehicles may decrease due to bumpy roads, potentially damaging the rescue equipment to a certain extent
We contribute an approach to reconstruct the terrain elevation map in real-time to extract the elevation values in advance based on the trajectory prediction. The significance of this project is that we use interdisciplinary knowledge to solve an engineering problem in the field of mechanical engineering, and the innovation of this manuscript belongs to Engineering Application Innovation
This paper introduces a method to reconstruct the terrain elevation maps in real-time to extract preview elevation information according to the predicted trajectory of a vehicle in advance

Summary

INTRODUCTION

While on route to rescue sites in remote areas with rough terrain, the chassis performance of emergency rescue vehicles may decrease due to bumpy roads, potentially damaging the rescue equipment to a certain extent. We contribute an approach to reconstruct the terrain elevation map in real-time to extract the elevation values in advance based on the trajectory prediction. We propose an approach to reconstruct the terrain map in real-time to extract the preview elevation information according to the trajectory prediction. LIDAR and inertial measurement unit (IMU) can be used to obtain high-accuracy point cloud and pose data to ensure the precision of the terrain elevation values This is the first application of its kind in which the uncertainty of the LIDAR model and vehicle motion are taken into consideration. We only care about the terrain map of vehicle direction in order to extract the elevation information based on the predicted trajectory The Gaussian probability distribution as Z ∼ N(hp, σh2P ) represents the estimated height of point P, hp represents the mean, and σh2P represents the variance of the elevation map

THE UNCERTAINTY ESTIMATION OF LIDAR MODEL IN THE PROCESS OF MEASUREMENT

TERRAIN ELEVATION MAP UPDATE

COVARIANCE COMPUTATION OF VEHICLE MOTION BASED ON GAUSSIAN RANDOM WALK MODEL

TRAJECTORY PREDICTION BASED ON A VEHICLE KINEMATICS MODEL

CONCLUSION