Fault Detection For Automatic Guided Vehicles Based on The Two-tower Model

Abstract

Automated Guided Vehicle (AGV) is one of the most important automation equipment in Automated Container Terminal (ACT), the normal operation of AGV equipment plays a vital role in maintaining high efficiency and quality operation of ACT. In this paper, We propose an end-to-end fault detection algorithm for AGV equipment that obtains both spatially regular and temporally dimensional features from the sensor data, which introduces a two-tower model structure for fault detection. The attention network is used to learn the key variables in the sensor data, and the LSTM network is combined to learn the temporal dimensional features in the data to construct a two-tower model for fault detection. In addition, we provide a real dataset from AGV at Qingdao port container terminals to evaluate the effectiveness of the algorithm. The experimental results show that this algorithm outperforms better than existing methods in terms of classification performance, and can achieve 98.83% accuracy.