From Robotaxis to L4 Autonomy: A Deep Dive Into NVIDIA Alpamayo 2 Super
Join us for a deep dive into Alpamayo 2 Super — our most capable model yet — as we walk through the release notebooks and show how it can power robotaxi and L4 autonomous vehicle pipelines at a fraction of the cost of existing solutions. We'll also share the latest AlpaSim updates, including new controller improvements, and give a full rundown of our two ongoing challenges — including how we've incorporated your feedback since the initial release. Attendees will learn: - How Alpamayo 2 Super combines reasoning, meta actions, and trajectory generation - How to use the released notebooks for trajectory generation, auto-labeling, VQA, and grounding - How NVIDIA is approaching distillation and Alpamayo Edge for onboard deployment - How AlpaSim uses reactive traffic and improved trajectory control for closed-loop evaluation - How AlpaGym enables targeted reinforcement-learning improvements - How the AlpaSim E2E Closed Loop Challenge evaluates autonomous-driving policies - How NVIDIA approaches safety, edge cases, and reasoning evaluation in AV development - How developers can start experimenting with Alpamayo 2 Super today ---------------------------------------------------------------------------- Chapters: 0:00 NVIDIA Alpamayo 2 Super Livestream Introduction 1:20 Meet the NVIDIA Autonomous Driving Research Team 2:45 Alpamayo 2 Super Capabilities 5:28 How Alpamayo Reasoning VLA Models Work 6:45 Alpamayo 2 Super Architecture and Benchmarks 9:00 Alpamayo 2 Super Notebook Demo 10:26 Trajectory Generation Demo 13:01 Meta Actions and Auto-Labeling 17:27 Visual Question Answering and Grounding 21:17 Alpamayo Edge and Distillation 22:22 Chain-of-Causation Auto-Labeling Challenges 25:00 ISO 26262, SOTIF, and AV Safety 27:26 AlpaSim Updates 28:06 Reactive Traffic Models 31:09 New SQP Tracking Controller 33:51 Faster Closed-Loop Evaluation 35:12 Closed-Loop RL With AlpaGym 38:05 Video Models as AlpaSim World Engines 40:06 AlpaSim E2E Closed Loop Challenge 43:34 Challenge Updates From Community Feedback 46:28 DGX Spark Challenge Prizes 47:05 Edge-Case Reasoning Benchmarks 49:10 Alpamayo Edge on NVIDIA DRIVE AGX Thor 49:51 Future-Frame Prediction and World Models 53:12 AlpaGym RL Rewards 54:39 3D Bounding Boxes and Post-Training 55:49 Prompt Injection and Robot Jailbreaking 57:37 Developer Resources and Next Steps 58:27 GTC Berlin and Automotive Sessions 1:01:27 Closing and Community Resources ---------------------------------------------------------------------------- List of Resources - Alpamayo 2 Super on Hugging Face - https://huggingface.co/nvidia/Alpamayo2-Super - AlpaSim E2E Closed Loop Challenge 2026 - https://huggingface.co/spaces/nvidia/AlpasimE2EClosedLoopChallenge2026 - Physical AI AV OOD Reasoning Challenge 2026 - https://huggingface.co/spaces/nvidia/PhysicalAI-AV-OOD-Reasoning-Challenge-2026 - Alpamayo Recipes GitHub - https://github.com/NVlabs/alpamayo-recipes Continue the discussion and ask follow-up questions in the NVIDIA Omniverse Discord thread: https://discord.com/channels/827959428476174346/1543113187610599525 ---------------------------------------------------------------------------- 📆 Check out the full calendar for all of our upcoming events → https://nvda.ws/3JqaWnA ---------------------------------------------------------------------------- ⬇️Get Started → https://nvda.ws/4cZAZO1 👀Explore OpenUSD → https://nvda.ws/3CeozBQ 👥Join the Community → https://nvda.ws/3ZMfc6e #NVIDIA #Alpamayo #Alpamayo2Super #AutonomousVehicles #AutonomousDriving #PhysicalAI #Robotaxi #L4Autonomy #AlpaSim #AlpaGym #NVIDIADRIVE #DRIVEAGX #ReinforcementLearning #Simulation #DeveloperTools