Sign in Subscribe
Machine learning

Patent Application for Parking Zone Mapping and Vehicle Localization Using Mixed-Domain Neural Network with Altitude Compensation

Strategic Insights

— 1 min read

The patent application, filed on January 5, 2024, by inventors Rachid Benmokhtar, Hung-Cheng Huang, Lihao Wang, and Xinhua Xiao, proposes a system and method for mapping parking zones and localizing vehicles using a mixed-domain neural network with altitude compensation. The system includes a plurality of image sensors mounted to a vehicle, generating raw image data, and an inertial measurement unit (IMU) generating movement data. The neural network is trained to generate image-domain data, bird's-eye-view (BEV) images, and BEV-domain data associated with parking landmarks.

Key Takeaways:

  • The patent application describes a method and system for mapping parking zones and localizing vehicles using a mixed-domain neural network with altitude compensation.
  • The system includes a plurality of image sensors and an IMU, generating movement data associated with the vehicle's movement.
  • The neural network is trained to generate image-domain data, BEV images, and BEV-domain data associated with parking landmarks.
  • The method includes receiving images from multiple cameras, executing a feature-detection model to generate image-domain data, generating BEV images and BEV-domain data, localizing the vehicle within the parking zone, and compensating the BEV-domain data and image-domain data based on the vehicle's movement data.
  • The system also includes generating tri-perspective view (TPV) images and TPV-domain data associated with parking landmarks.
  • The method includes compensating the TPV-domain data and the mapping of the parking zone is further based on the compensated TPV-domain data.
  • The system can be used in various applications, including autonomous or semi-autonomous driving, parking assistance, and vehicle-to-everything (V2X) communications.

Statistics:

  • 4 inventors (R. Benmokhtar, H.-C. Huang, L. Wang, and X. Xiao) filed the patent application on January 5, 2024.
  • 20 claims are included in the patent application.
  • The system includes 1+ image sensors and 1 IMU generating movement data.
  • 2 types of neural networks are mentioned (feature-detection model and mixed-domain neural network with altitude compensation).
  • 2 types of images are generated (BEV and TPV images).

Sources:

  • BENMOKHTAR, Rachid; HUANG, Hung-Cheng; WANG, Lihao; XIAO, Xinhua. Methods And Systems For Parking Zone Mapping And Vehicle Localization Using Mixed-Domain Neural Network With Altitude Compensation. U.S. Patent Application Number 20250222950, filed January 5, 2024 and posted July 10, 2025. Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(20250222950)&db=US-PGPUB&type=ids

Read more