DARPA wants to move faster on autonomous off-road UGV simulation

DARPA has awarded an Intel-led team a four-year contract to develop advanced simulation solutions that accelerate the R&D process for off-road autonomous ground vehicles.

The two-phase Robotic Autonomy in Complex Environments with Resiliency – Simulation (RACER-Sim) programme ‘aims to create the next generation of off-road simulation platforms to significantly reduce the development cost and bridge the gap between simulation and the real world’, Intel noted in a 26 April statement.

Its team includes the Barcelona, Spain-based Computer Vision Center and the University of Texas at Austin.

In its first phase, RACER-Sim will focus on creating new simulation platforms and map generation tools that mimic complex off-road environments with the highest accuracy ‘at scales never seen before’ covering more than 100,000 square miles (260,000km²), Intel noted in its statement.

In phase two, researchers will seek to catalyse the R&D process by implementing new algorithms without the use of a physical robot. This will be followed by new techniques that enable autonomous UGVs to be trained directly in simulation.

There is currently a gap between on-road and off-road deployment in the context of autonomous driving.

Few of the simulation environments available today are optimised for off-road autonomy development at scale and speed, and real-world demonstrations remain the primary method to verify system performance.