NVIDIA Launches AI Models to Make Self-Driving Vehicles Think Like Humans

SadaNews - NVIDIA launched "Alpamayo," a new family of open-source AI models, simulation tools, and datasets for training physical robots and vehicles at the Consumer Electronics Show (CES 2026), designed to help self-driving vehicles think in complex situations while driving.

Jensen Huang, the CEO of NVIDIA, stated, "The moment for embodied AI similar to ChatGPT has arrived - when machines start to understand, analyze, and act in the real world."

Huang added, "Alpamayo brings reasoning capabilities to self-driving vehicles, enabling them to consider rare scenarios, drive safely in complex environments, and explain their driving decisions," according to a report from TechCrunch that was seen by Al Arabiya Business.

The "Alpamayo 1" model is the foundation of NVIDIA's new model family, a reasoning-based language model with a chain of thoughts consisting of 10 billion parameters, allowing the self-driving vehicle to think in a manner closer to humans, enabling it to find solutions for complex exceptional cases - such as how to navigate a malfunctioning traffic signal at a busy intersection - without prior experience.

Ali Kani, Vice President of Automotive at NVIDIA, said during a press conference on Monday, "(The system) does this by breaking problems into steps, considering all possibilities, and then choosing the safest path."

As Huang explained during his keynote on Monday, "Alpamayo's role is not limited to receiving sensor data and activating the steering wheel, brakes, and acceleration; it also thinks about the action it is about to take. It tells you the action it will take, and the reasons that led it to this action. Then, of course, it considers the path."

The core code for the "Alpamayo 1" model is available on the "Hugging Face" platform. Developers can enhance the model's performance to produce smaller and faster versions for vehicle development, or use it to train simpler driving systems, or build tools on top of it such as automatic classification systems that automatically classify video data, or evaluation tools to validate the car's decisions.

Kani stated: "They can also use Cosmos to generate synthetic data, and then train and test self-driving applications based on Alpamayo using a mix of real and synthetic data."

"Cosmos" is NVIDIA's trademark for generative world models, which are AI systems that create a representation of the physical environment to enable predictions and actions.