Share Email Print

Proceedings Paper

A learning-based autonomous driver: emulate human driver's intelligence in low-speed car following
Author(s): Junqing Wei; John M. Dolan; Bakhtiar Litkouhi
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

In this paper, an offline learning mechanism based on the genetic algorithm is proposed for autonomous vehicles to emulate human driver behaviors. The autonomous driving ability is implemented based on a Prediction- and Cost function-Based algorithm (PCB). PCB is designed to emulate a human driver's decision process, which is modeled as traffic scenario prediction and evaluation. This paper focuses on using a learning algorithm to optimize PCB with very limited training data, so that PCB can have the ability to predict and evaluate traffic scenarios similarly to human drivers. 80 seconds of human driving data was collected in low-speed (< 30miles/h) car-following scenarios. In the low-speed car-following tests, PCB was able to perform more human-like carfollowing after learning. A more general 120 kilometer-long simulation showed that PCB performs robustly even in scenarios that are not part of the training set.

Paper Details

Date Published: 7 May 2010
PDF: 12 pages
Proc. SPIE 7693, Unattended Ground, Sea, and Air Sensor Technologies and Applications XII, 76930L (7 May 2010); doi: 10.1117/12.852413
Show Author Affiliations
Junqing Wei, Carnegie Mellon Univ. (United States)
John M. Dolan, Carnegie Mellon Univ. (United States)
The Robotics Institute, Carnegie Mellon Univ. (United States)
Bakhtiar Litkouhi, GM-CMU Autonomous Driving Collaborative Research Lab., General Motors Corp. (United States)

Published in SPIE Proceedings Vol. 7693:
Unattended Ground, Sea, and Air Sensor Technologies and Applications XII
Edward M. Carapezza, Editor(s)

© SPIE. Terms of Use
Back to Top