Chip-maker Nvidia has been awarded a $20 million Defense Advanced Research Projects Agency (DARPA) contract to develop embedded processors that could be used to process data collected from sensors in autonomous vehicles.
The five-year contract, which has been awarded as part of DARPA's Power Efficiency Revolution For Embedded Computing Technologies (PERFECT) program, has the goal of developing processors that have 75-times the processing power of current embedded systems. These processors would find their way into autonomous air and ground vehicles to enable real-time surveillance analysis.
Existing embedded processors can deliver around one gigaflops of performance per-watt. The Nvidia program, dubbed Project Osprey, will look to develop power efficient architectures and programming systems that should lead to processors based on 7-nanometer architectures -- four-times smaller than the current 28-nanometer technology -- capable of delivering an incredible 75 gigaflops per-watt.
"The technologies developed with this program can transform the capabilities of embedded systems, making autonomous vehicles more practical and intelligent," said Steve Keckler, Nvidia senior director of architecture research. "This research will help Nvidia continue to advance mobile computing for both government and consumer applications."
The research will be carried out alongside academics from the University of Utah and the University of Virginia.