Real-Time Vehicle Positioning and Mapping Using Graph Optimization.

Anweshan Das,Jos Elfring,Gijs Dubbelman

doi:10.3390/s21082815

Anweshan Das, Jos Elfring + Show 1 more

Open Access

https://doi.org/10.3390/s21082815

Copy DOI

Abstract

In this work, we propose and evaluate a pose-graph optimization-based real-time multi-sensor fusion framework for vehicle positioning using low-cost automotive-grade sensors. Pose-graphs can model multiple absolute and relative vehicle positioning sensor measurements and can be optimized using nonlinear techniques. We model pose-graphs using measurements from a precise stereo camera-based visual odometry system, a robust odometry system using the in-vehicle velocity and yaw-rate sensor, and an automotive-grade GNSS receiver. Our evaluation is based on a dataset with 180 km of vehicle trajectories recorded in highway, urban, and rural areas, accompanied by postprocessed Real-Time Kinematic GNSS as ground truth. We compare the architecture’s performance with (i) vehicle odometry and GNSS fusion and (ii) stereo visual odometry, vehicle odometry, and GNSS fusion; for offline and real-time optimization strategies. The results exhibit a 20.86% reduction in the localization error’s standard deviation and a significant reduction in outliers when compared with automotive-grade GNSS receivers.

Highlights

Autonomous driving technologies are evolving rapidly with the ultimate goal of developing a safe and reliable, fully autonomous vehicle, i.e., SAE level 4 and eventually level 5 [1]
In Section 7.3.4, we analyze the performance of the proposed framework with (i) vehicle odometry and Global Navigation Satellite System (GNSS) fusion and (ii) stereo-visual odometry (SVO), vehicle odometry, and GNSS fusion, using the selected pose-graph model, the batch size, and the optimization window size
We have proposed and evaluated a pose-graph based real-time multi-sensor fusion framework for vehicle positioning and mapping using a stereo camera, vehicle’s yaw-rate, velocity sensor, and a GNSS receiver

Summary

Introduction

Autonomous driving technologies are evolving rapidly with the ultimate goal of developing a safe and reliable, fully autonomous vehicle, i.e., SAE level 4 and eventually level 5 [1]. A real-time, accurate, and robust positioning system is the backbone of a fully autonomous vehicle and many Advanced Driver Assistance Systems (ADAS). It is the basis for environment perception, path planning, and autonomous decision making. Global Navigation Satellite System (GNSS) is most widely used for vehicle positioning (GPS, GLONASS, Galileo, BeiDou are all examples of GNSS systems) These systems are not always reliable as they are dependent on satellite visibility. The state-ofthe-art INS uses tactical-grade IMUs with Real-time kinematics (RTK) GNSS receivers to estimate positions accurately. Their data are often postprocessed to achieve centimeter-level accuracy. This article focuses on improving the availability, accuracy, and reliability of the vehicle positioning system fusing low-cost automotive-grade sensor data

Objectives

Results

Conclusion