June 5, 2014 · 6:00 PM
This location is shown only to members
You asked for technical, we're giving you technical! Come learn how Paul has researched pushing the limits of an autonomous vehicle and competing with professional drivers on a race course.
Again, food starts at 6pm, talk starts at 6:30pm. See you there!
Automated vehicles can potentially improve the safety of the average driver in emergency situations. Many accidents occur because the car has reached the limits of its capability such as in hazardous weather conditions. Professional racecar drivers routinely operate at the limits of handling to win races, so a unique way to provide safety systems for future automated vehicles is to emulate the skills of professional drivers. The presentation will discuss how a family of simple curves can be used to create trajectories that incorporate professional driving techniques. Experimental results of the resulting trajectory was implemented on an automated vehicle demonstrate performance that surpasses the average driver. This path planning structure offers an approach that can be generalized to many different safety critical situations.
Paul Theodosis is a PhD student at Stanford University working on autonomous racecars. His research centers on path planning at the limits of handling, dynamic modeling to problems in nonlinear control, estimation and diagnostics. He was part of the autonomous Audi TTs project that raced up Pikes Peak in 2010. He now tests the autonomous Audi at Thunderhill Raceway Park in Willows, Ca. At this racetrack, he works on improving the car’s lap time to beat the best human drivers and developing real-time trajectory modification to avoid danger. He currently works in the Dynamic Design Lab with Chris Gerdes as his advisor.