A two-level path planning and monitoring architecture for real-world autonomous driving systems

In the past two decades, the safety of autonomous vehicles has drawn in-creasing attention from both academic researchers and industrial experts.This work presents a novel two-level path planning and monitoring architecture to promote safety in autonomous driving. We implemented a HybridA* offline path planner that can generate paths for the vehicle to navigate through static obstacles. The path is guaranteed to be achievable by the vehicle. Also an online monitor is developed using reachability analysis. It simulates the movements of the vehicle based on the vehicle’s physical model and checks the safety of interaction between the vehicle and surrounding dynamic environment in near-real-time (∼0.1 s). The combined architecture, as part of the GEM project, has been tested in both real world and simulation environments. In the experiments, the output path from the path planner is always achievable by the vehicle without colliding with static obstacles, and the online monitor shows high fidelity in predictions while the running time is only slightly longer than that of a simple baseline algorithm.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2020

Author dc:creator

• Liu, Tianqi

Contributors dc:contributor

• Mitra, Sayan

• autonomous driving, reachability analysis

Statement dc:rights

• Copyright 2020 Tianqi Liu

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/108162

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/108162