Video Based Localization Using V2X, Machine Learning, and Blockchain Storage

Global Positioning Systems (GPS) are critical to enable Vehicle-to-Everything (V2X) safety applications for smart cities. To uphold the reliability of V2X applications when GPS is not available, we propose a trusted video-based localization system using incremental machine learning and a private blockchain to enable GPS estimation. In addition to a blockchain consensus algorithm, we propose three new SAE J2735 messages for enabling machine learning based video-positioning estimation over V2X. We validate the proposed approach with a computer vision classifier that provides 2 meters of estimation error.