NVIDIA Alpamayo 2 Super
NVIDIA Alpamayo 2 Super is a frontier open model for robotaxis and autonomous vehicles, built to reason through rare, complex driving situations and produce decisions developers can inspect, validate, and trust. Based on NVIDIA Cosmos 3 Super Reasoner and post-trained with reinforcement learning, it combines commercial openness with multitask capabilities for autonomous driving. The model reasons across full-surround camera coverage, fusing front, side, and rear views to understand lane changes, merges, unprotected turns, and complex intersections. For each scenario, it can generate a planned vehicle trajectory, a chain-of-causation trace explaining the decision, a meta-action such as yielding or stopping, reasoning auto-labels for training and validation, and visual question-answering responses grounded in specific image regions. These linked outputs make it easier to connect what the model observed with the action it selected.
Learn more
Waymo
Waymo is an autonomous driving technology company that develops self-driving vehicles and operates fully driverless transportation services. Originally created as Google’s self-driving car project in 2009, the company later became an independent subsidiary of Alphabet with the goal of making transportation safer, more accessible, and more efficient through autonomous mobility. Its core technology, known as the Waymo Driver, combines artificial intelligence, high-resolution cameras, radar, lidar sensors, and detailed digital maps to allow vehicles to perceive their surroundings and navigate roads without human intervention. It continuously analyzes traffic signals, pedestrians, other vehicles, and road conditions to determine safe driving actions in real time. Before operating in a new area, Waymo vehicles map roads in extreme detail, identifying lane markings, signs, and intersections, and then combine this information with real-time sensor data to maintain precise positioning.
Learn more
Helm.ai
We license AI software throughout the L2-L4 autonomous driving stack, perception, intent modeling, path planning, and vehicle control. Highest accuracy perception and intent prediction, leading to safer autonomous driving systems. Unsupervised learning and mathematical modeling, instead of supervised learning, allow learning from huge datasets. Our technologies are up to several orders of magnitude more capital-efficient, enabling much lower cost of development. Helm.ai full scene vision-based semantic segmentation fused with Lidar SLAM output from Ouster. L2+ autonomous driving with Helm.ai across highways 280 to 92 to 101, lane-keeping + ACC lane changes. Helm.ai pedestrian segmentation, with key-point prediction. Pedestrian segmentation and keypoint detection. Rain lane detection corner cases and Lidar-vision fusion. Full scene semantic segmentation, botts dots, and faded lane markings.
Learn more
openpilot
openpilot is open source software built to improve upon the existing driver assistance in most new cars on the road today. Tesla Autopilot like functionality for your Toyota, Honda, and other top brands. While engaged, openpilot includes camera based driver monitoring that works both day and night to alert the driver when their eyes are not on the road ahead.
Learn more
