
NVIDIA · 2026-11-10 16:00 UTC
Running AV Reasoning Models Within Onboard Compute and Latency Limits
NVIDIA Alpamayo 2 Edge is an adaptive reasoning vision-language-action model built on a 2B Nemotron-based backbone and designed for real-time operation on NVIDIA DRIVE AGX Thor. It generates Chain-of-Causation reasoning and future driving trajectories from surround-camera observations, temporal context, and navigation inputs, reasoning only when needed to conserve compute. Closed-loop reinforcement learning post-training in NVIDIA AlpaGym improves closed-loop driving performance by 48.9% over the best checkpoint without closed-loop RL.
Loading the story on Tree News…