放开自动驾驶汽车安全竞争研究.pdf

Beyond Autonomous Cars; Open Autonomous Vehicle Safety Competitions Mike Cannon Boyd Wilson Clemson University Omnibond Abstract By applying open collaborative concepts, to autonomous vehicle research, we could further advance the safety in “self driving” vehicles. The vision is to have an open autonomous vehicle safety competition, in which a vehicle is required to maneuver a difficult course, erroring on the side of least bodily injury. The specifics of this session are to see if there is interest in autonomous vehicle safety competitions. There will be a demonstration of TrafficVision, an existing “off auto” sensor technology. The session will end with thoughts on what an open autonomous vehicle safety competition venue would consist of, specifically seeking input from the community. Audience: Automotive Enthusiasts, Robotic / autonomous vehicle engineers, Robotic / autonomous software developers The ultimate benefits of the this would produce safer transportation system(s). Decrease the time to have working autonomous vehicles on city streets. Create new collaborative environments between the auto industry and universities focused on open collaboration for safety. How to get Autonomous Vehicles Accepted How I got Interested Discussions at Conferences: Linuxcon Europe, SCaLE Automotive Linux Summit (an odd mix?) Mark Hinkle’s “We Won! What’s Next?” Presentation; /socializedsoftware /markhinkle/scale12x-whyopensource Panel: Autonomous Vehicles and Open Source Collision or Cooperation /sites/events/files/slides/edinburgh %20ALS%20panel.pdf Benchmarks Specifically, IT professionals benchmark systems with a tool to simulate real world scenarios. - io generator for storage or network performance. Speeds up vetting issues, less wasting of other peoples time, fun to do this type of work. Benchmarks can be applied to any topic. Safety Requirements vs. Benchmarks NHTSA defines the minim