Nvidia patents radar object detection using self-taught AI
Nvidia describes a method where an AI model learns to detect objects by comparing raw radar data with its own reconstructed version. The system estimates object properties like velocity and uses the difference between the original and reconstructed data to improve its accuracy without needing labeled examples.
In the filing’s own words
A machine learning (ML) model may be trained using a self-supervised approach. Training sensor data may be provided to the ML model, and the ML model may estimate one or more object properties based on the training sensor data. Reconstructed sensor data may be generated based at least on the estimated object properties. The reconstructed sensor data may be of a same data type as the training sensor data. One or more parameters of the ML model may be updated based at least on a comparison between the training sensor data and the reconstructed sensor data. A machine may obtain sensor data using one or more sensors. The machine may determine one or more properties corresponding to an object based at least on the ML model processing the sensor data. The machine may perform one or more operations based at least on the determined properties.
What exactly they claim (claim 1)
1. A method comprising: performing, by a machine, one or more operations based at least on an output of a machine learning (ML) model as generated using RADAR data, wherein the ML model was trained, at least in part, by: comparing training RADAR data to reconstructed RADAR data, wherein the reconstructed RADAR data was reconstructed based at least on one or more previously estimated velocities previously estimated based at least on the ML model processing the training RADAR data, the reconstructed RADAR data being of a same data type as the training RADAR data.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- OBJECT DETECTION
- Publication number
- US20260328905A1
- Application number
- 19726152
- Filed
- Jun 30, 2026
- Published
- Oct 22, 2026
- Inventors
- 3
- Claims
- 20
Classified under
G01S 7/417 Physics & computingG01S 13/58 Physics & computingG01S 13/931 Physics & computingMore from Nvidia in Cars & transport
- Nvidia patents robotaxi system with AI dispatcher and virtual rails Sep 15, 2026
- Nvidia patents car drowsiness check that uses your body clock Oct 13, 2026
- Nvidia patents 3D camera system to measure passengers in vehicles Oct 13, 2026
- Nvidia patents cruise control that predicts driver's next move Oct 20, 2026
- Nvidia patents dual-brain system for self-driving cars Sep 17, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.