Aurora’s Virtual Testing Suite has been an important part of our company and today we’re accelerating our simulation efforts by welcoming the team from Colrspace – creative technology start-up made up of Pixar veterans behind the computer-generated imagery (CGI) magic of iconic movie series like “Toy Story” and “Cars.”
Colrspace’s team bridges state-of-the-art computer graphics and machine learning with a data-driven approach to reconstructing 3D objects and materials, deployable in real-world environments. As we continue to leverage and expand our Virtual Testing Suite, Colrspace's technology will bring scalability and increased accuracy to the high-fidelity virtual worlds that underpin our unique sensor simulation capabilities. Ultimately, Colrspace's team and technology will enable us to move even faster in developing simulation and machine learning tools, accelerating our progress toward delivering the Aurora Driver.
With our industry-leading Virtual Testing Suite, Aurora runs millions of simulations every day. This allows us to train and evaluate the Aurora Driver’s software stack across a vast range of scenarios and driving conditions, finding edge cases and catching errors early, well before the software is loaded onto vehicles. Ultimately, simulation testing drives the development of the Aurora Driver. It is the quickest, and safest way to train and test our self-driving technology – estimated to be equivalent to more than 50,000 trucks driving continuously. This, combined with thoughtful on-road testing, will allow us to deliver the Aurora Driver safely and quickly at scale.
As for Colrspace, its innovation is in Protocolr, which is based on an input image and infers texture and other material properties of an object. Key to the process is a neural network that models the processing of the camera pipeline, and couples this with a differentiable image renderer to enable “inverse rendering,” which computes the 3D scene and materials that would produce an image identical to the input photo. Aurora believes this technology will provide a unique advantage in building virtual worlds that are almost indistinguishable from the real one. This is critical because the more realistic the virtual world, the more effective the testing can be.