Truck Automation and Safety (TAS)

1.Function Introduction

Install lasers and vision in the on-site yard crane and truck operation area, and install edge AI servers in the electrical room. Use artificial intelligence algorithms to fuse and analyze laser point clouds and visual images to achieve precise positioning and perception of trucks. Through a set of software and hardware, provide truck guidance, truck anti lifting, box bottom lock recognition, anti smashing front, and other four functions in one. Each function has dual technology or dual channel fusion complementarity to ensure stable operation around the clock.

2.Technical indicators

Analysis and precise identification time for trucks and containers:<100 milliseconds;
Container positioning accuracy:<± 30 millimeters; Truck positioning accuracy:<50 millimeters;
Container crane lifting recognition distance: generally<40 centimeters;
Target recognition accuracy ≥ 99.9%, false alarm rate: ≤ 0.1%;
The recognition rate of unresolved container bottom locks is ≥ 95%; The accuracy of front position recognition is ≥ 99.99%;
Supporting container sizes: single 20 feet, double 20 feet, 40 feet, or 45 feet;

3.System function

Truck Positioning, Anti-Truck Lifting, Twist-lock Identification, and Anti-Truck Cab Collision

The controller sends a heartbeat signal to the PLC, which determines whether the system is faulty;
Once a truck accident occurs, the remote control panel cannot resolve the fault by shutting down the system, but must be checked by engineering personnel and resolved through fault reset;
The system provides a video data interface, seamlessly connecting the camera's image with the CCTV system, which helps remote control operators to observe and confirm the real-time operation status of RMG.

4.System advantages

This system adopts a laser and vision fusion (hereinafter referred to as laser vision fusion) scheme, using the SmartDONN artificial intelligence algorithm of DONN Technology to analyze and process laser point cloud and visual image data, quickly and accurately identify key targets such as frames, containers, lock holes, wheels, etc., accurately calculate the position and posture of trucks (trailers) and containers, and determine whether there are any abnormal situations. The system utilizes various advanced algorithms, including laser point cloud modeling, RGB image analysis, neural network learning, and high-speed hash classification.