AV 3.0

AV 3.0 is a data augmentation scheme for training improved AV 2.0 end-to-end networks. Here we are safely, smoothly, and realistically driving all the cars in these videos, including the ego, on our traffic light controlled HD maps. Rendered using NVIDIA COSMOS. Interested in learning how we can help you dramatically and quickly improve your end-to-end AV stack?
Learn More

AV 2.0

AV 2.0 refers to the end-the approach famously taken by Tesla and now NVIDIA, Wayve, Zenseact, and GM among others. Even with Tesla-scale fleet and training operations AV 2.0 systems still struggle to perform. See the FSD tracker to see how far this approach still has to go. AV 3.0 is the answer.
Learn More

AV 1.0

AV 1.0 refers to the modular AV/ADAS architectures employed famously by Waymo but also Mercedes, May, Latitude, and others. Inverted AI's fully-validated low-TOPS realtime prediction and planning modules have successfully driven on road. Interested in integrating them into your onboard stack? Or more comprehensively validating your post-perception stack?
OnboardPost Perception V&V