Crystal edge AI compute for autonomous trucks

  • March 9, 2021
  • William Payne

Self driving truck manufacturer TuSimple is working with edge computing firm Crystal Group and robotics developers to improve autonomous self driving AI technology for trucks and heavy construction machinery. As part of this development, Crystal Group has introduced a range of rugged autonomous vehicle compute solutions in edge based systems.

The systems are built to provide seamless real-time situational awareness and operation regardless of ever-changing conditions. Crystal Group’s autonomous vehicle compute (AVC) solutions are designed to combine proprietary ruggedization techniques and 12-V DC, ATX-compliant server with NVIDIA’s robust GPUs. 

Using a power supply that interfaces directly with the vehicle’s battery, these servers are designed to eliminate the need for a separate power converter while delivering 1,500 or 2,000 watts of power to interpret and act upon the massive inputs captured by the vehicle’s network of sensors, radar and cameras. 

An optional liquid cooling system outputs the waste heat to prevent throttling due to the significant heat generated by the power-hungry compute requirements.

“Equipping these substantial machines for safe, successful driverless operation with little or no human intervention requires us to understand the operational requirements in tandem with the range of environmental conditions to which they may be exposed,” said Micah Snodgrass, technical director at Crystal Group. “Ongoing, candid collaboration with TuSimple resulted in a custom solution with enhanced processing power, thermal management and shock absorption that not only met their functional requirements, but also raised industry benchmarks for safety, efficiency and reliability.”