Vehicle Path Following and Rollover Prevention Using Previewed State Information
Open Access
Author:
Stankiewicz, Paul Geoffrey
Area of Honors:
Mechanical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Dr. Sean N Brennan, Thesis Supervisor Henry Joseph Sommer III, Thesis Honors Advisor
Keywords:
path following rollover preview control
Abstract:
The research in this thesis focuses on investigating methods of vehicle path
following and rollover prevention with application towards autonomous vehicles.
Statistics show that although rollover only occurs in 2.2% of total highway crashes, it
accounts for 10.7% of total fatalities. Autonomous vehicles must be able to remain
within the bounds of the road, while also preventing rollover during emergency
situations. Vehicle path following is a mature problem and has been investigated
several ways, one of which will be used and evaluated in this research. There are also
several dynamic rollover metrics in use that measure a vehicle's rollover propensity
under specified conditions. However, in order to prevent a rollover event from
occurring, it is necessary to predict a vehicle's rollover propensity in the future. This research uses a novel vehicle rollover metric, called the zero-moment point (ZMP), to
predict the vehicle's rollover propensity. Comparing different amounts of preview, the
results show that short-range predictions - as little as 0.75 seconds ahead of the vehicle
- are sufficient to prevent nearly all dynamics-induced rollovers in typical shoulders
and medians.
