Audi has officially partnered up its self-driving car unit, Autonomous Intelligent Driving, with an Israeli AI startup by the name of Cognata. Specifically, AID has chosen Cognata to help put its fleet of autonomous vehicles through simulations. This means that Cognata and its simulation solution will be responsible for helping AID’s vehicles build their collective driving style and knowledge base before they’re able to hit the streets for testing in the real world. As part of the multi-year deal, Cognata will not only provide the simulation, but will manage the cloud and serving technologies behind the simulation, manage connections between vehicles, and provide support to Audio throughout the life cycles of all vehicles it puts through its simulation.
Cognata’s simulation uses AI technology, along with deep learning and computer vision, to make convincing models of real cities that present many of the same challenges as the real deal. Traffic and miscellaneous hazards, such as pedestrians and construction, are faithfully reproduced based on data that Cognata has obtained. Not only will the simulations present the cars with realistic environments, those environments will produce emulated responses from the cars’ many sensors, meaning that driving in the simulation will be a near-perfect approximation of real-world testing. Since AID’s vehicles run in a gestalt, like most self-driving cars, they will not only be running the simulation together, but will be putting their collective computer power and knowledge together when it comes time for Cognata to run an assessment, as well.
AID, owned by Aldi, is somewhat unique in the self-driving car game in a couple of ways; the firm is developing its own solution from the ground up, something that is normally relegated to the bigger players like Waymo and Uber, and once it’s done, it won’t simply keep the platform to Audi cars, but will create a universal platform of sorts that any compatible vehicle can use. Cognata, as well, is somewhat unique; it specializes in creating simulations and tests for autonomous driving systems, but is using deep learning and AI technology in a potentially disruptive way, allowing its simulations to learn right alongside the cars that they’re teaching.