US Patent 11475680 Cargo sensor system implemented using neural network

A cargo sensor system incorporates an optical cargo sensor to supply images to a convolutional neural network. In preferred embodiments, the neural network is implemented using a processor in a sensor module, and is trained by a machine learning system to determine the load state of the cargo container. Some embodiments also include a secondary sensor, such as a laser-ranging Time-of-Flight (ToF) sensor, that verifies the cargo reading determined by the optical cargo sensor.