Intel develops advanced solutions and simulations for off-road vehicles to help facilitate the transition from virtual to real-world environments. Company; Together with the Barcelona, Spain-based Computer Vision Center and UT Austin, DARPA is advancing RACER’s simulation program in autonomous off-road vehicle simulation and the transfer of learning from simulation to the real world.
Defense Advanced Research Projects Agency (DARPA), Intel Laboratories as well as its collaborators Computer Vision Center based in Barcelona, Spain and Intel Federal LLC, supported by the University of Texas at Austin presented the opportunity to develop advanced simulation solutions for autonomous off-road vehicles. Robotic Autonomy – Simulation in Complex Environments with Flexibility (RACER-Sim) program aims to create next-generation terrain simulation platforms to significantly reduce development cost and bridge the gap between simulation and real world.
Speaking on the subject, German Ros, Director of the Autonomous Vehicles Lab at Intel Laboratories, said, “Intel Labs is already making progress in developing autonomous vehicle simulation through various projects such as the CARLA simulator. We are proud to join RACER-Sim to contribute to exploring new horizons in the field of off-road robotics and autonomous vehicle robots. We have assembled a team of renowned experts from the Computer Vision Center and UT Austin to create a versatile and open platform that will accelerate progress in terrain robodollars for any environment and condition.” said.
In the context of autonomous driving, the difference between on-road and off-road distributions is still very significant. While there are many simulation environments available today, very few of them are optimized for improving terrain autonomy at high scale and speed. In addition, real-world demos still serve as the primary method of verifying system performance.
Autonomous off-road vehicles have to cope with significant challenges such as the absence of road networks and extreme soil conditions covered with stones and vegetation of all kinds. Such extreme conditions increase the cost and slow down the development and testing processes. The RACER-Sim program aims to overcome this problem by providing advanced simulation technologies to develop and test solutions and shorten the deployment time and validation of AI-powered autonomous systems.
RACER-Sim consists of two phases, lasting 48 months in total, aiming to accelerate the entire research and development process in the design of autonomous off-road vehicles. In phase one, Intel is focusing on building new simulation platforms and mapping tools that emulate complex terrain environments at unprecedented scales with the highest accuracy (e.g. physical features, sensor modeling, terrain complexity, etc.). Building high-scale simulation environments using traditional methods requires significant resources, and this is one of the biggest challenges in simulation workflows. Intel Labs’ simulation platform will enable the customization of maps of the future, including creating massive environments spanning more than 100,000 square miles with just a few clicks.
In the second phase, Intel Labs will work with RACER collaborators to accelerate research and development by applying new algorithms without using a physical robot. Teams will then validate the robot’s performance in simulation, saving significant time and resources. The second phase will also include the development of new sim2real (from simulation to reality) techniques. These techniques aim to provide skills by training the robot in simulation and then transfer these skills to a real robotic system. In this way, the training of autonomous off-road vehicles can be done directly in the simulation.
Intel expects these new simulation tools to significantly improve the development of autonomous systems by using virtual tesdollarser, which reduces the risks, cost, and delays associated with traditional test and validation protocols. In the future, the simulation platform will go beyond validation to create AI models that are ready to be applied in the real world.
0 Comments