Team Cornell and the 2007 Urban Challenge: Research, Results and Next Steps

Google Tech Talks
January, 17 2008

ABSTRACT

Team Cornell was one of six teams to complete the 2007 DARPA Urban Challenge, completing over 55 miles of autonomous driving in an urban environment in approximately seven hours, including competition stops. The competition included many urban driving scenarios such as staying in a lane, merging into traffic, passing, intersections, parking, and even robot-robot interaction. Team Cornell designed and built a vehicle around technological innovations in vehicle automation, a real time UDP based data distribution system, tightly coupled pose estimation, scene estimation including localization within an urban environment and tracking all obstacles with a fusion of laser, radar and vision sensors, and hierarchical intelligent planning. Team Cornell's vehicle was designed to drive "human-like" with smooth, intelligent behaviors, even in the presence of a vast array of uncertainties. The systematic approach taken by Team Cornell led to an innovative, robust solution to the complex problem proposed in the 2007 DARPA Urban Challenge. This seminar will present the key technologies, semi-final and final results, and plans for future research.

Speaker: Dan Huttenlocher
Dan Huttenlocher is the John P. and Rilla Neafsey Professor of Computing, Information Science and Business at Cornell University, where he holds a joint appointment in the Computer Science Department and the Johnson Graduate School of Management. His research interests are in computer vision, social and information networks, collaboration tools, geometric algorithms and financial trading systems. He has been recognized for his research and teaching contributions on several occasions, including being named an NSF Presidential Young Investigator, New York State Professor of the Year and a Fellow of the ACM. In addition to academic posts he has been chief technical officer of Intelligent Markets, a provider of advanced trading systems on Wall Street, and spent more than ten years at Xerox PARC directing work that led to the ISO JBIG2 image-compression standard.

Speaker: Mark Campbell
Mark Campbell is an Associate Professor in the Sibley School of Mechanical and Aerospace Engineering at Cornell University. His research interests are in the areas of autonomous systems, probabilistic models of human decision making, nonlinear estimation theory, cooperative vehicle control and estimation, and sensor fusion. He has been recognized from NASA for his modeling and control work on the Middeck Active Control Experiment, flown on STS-67 in 1995. He received best paper awards from the AIAA and Frontiers in Education conference, and teaching awards Cornell, University of Washington, and the ASEE. He was also an Australian Research Council International Fellowship in 2006 while on sabbatical at the University of Sydney. He is an Associate Fellow of the AIAA, an Associate Director of the AACC board, and member of the AIAA GNC Technical Committee, and is active in both IEEE and ASEE.

Category:

People & Blogs

Tags:

• google
• techtalks
• techtalk
• engedu
• talk
• talks
• googletechtalks
• education

[bertilhatt]

Raman's question was about *replacing* the red lights, right? I'm assuming a functionning full-scale system would need more communication between vehicules, through Wifi e.g., and fix elements on dangerous points.

[zassounotsukushi]

yeah, they sort of moved them all to youtube. To be honest, I kind of liked having them in Google video better - larger basic screen size and stuff.

Good update on the DARPA challenge though. If these things aren't driving me around in 3 years I'm going to be disappointed!

[josecampanello]

I've subscribed to the RSS feed using Google Reader.

All but one (from Jan 8) of the las 10 items in the feed say the video is unavailable, but then, on youtube, the videos are ALL available.