Open AccessA Moving Object Detection and Predictive Control Algorithm Based on Deep Learning
Author(s) -
Yan He,
Xiuxian Li,
Hongfei Nie
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2002/1/012070
Subject(s) - computer science , path (computing) , algorithm , trajectory , interference (communication) , artificial intelligence , stability (learning theory) , object detection , computer vision , pattern recognition (psychology) , machine learning , astronomy , programming language , computer network , channel (broadcasting) , physics
In traffic scene, a low detection rate of dynamic target is caused by interference features of background area and fast speed of detected moving target. In this paper, an optimal target detection and prediction algorithm is proposed. Firstly, the algorithm of real-time motion parameters (speed, direction, etc.) detection of moving objects (vehicles) based on the depth theory is studied. The original prediction problem is transformed into the problem of automatic updating rate of uncertain parameters and the problem of minimizing the maximum path distance. Secondly, by on-line estimation of the automatic update rate, the proposed trajectory generated is guaranteed to be the minimum length. The allocation strategy minimizes the maximum distance traveled in the collision free. Then, the stability of the estimation error system is guaranteed base on deep learning algorithm. Finally, the simulation shows that the proposed algorithm can be used for moving target detection and path prediction accurately. The performance of the control system is improved.