An outdoor sheltered parking space detection system combined with drone and GNSS-RTK technology

Aiming at the shortcomings of the existing parking space detection system or detection technology, proposed and preliminarily realized a kind of quad-roder UAV as a mobile platform, supplemented by GNSS-RTK high-precision satellite positioning, and image recognition technology's parking space detection system. An artificial neural network is used in the system to complete the real-time detection of the parking space image in the airborne computer, and then only need to send the detection results back to the ground for announce. Can be used for sheltered outdoor parking lots, and when used in parking lots of the same size, the deployment cost is the existing system's 1/4~ 1/12. Due to the natural high noise immunity of neural networks, and the relative fixity of the use of the scene, with the support of GNSS-RTK technology, it is passed Self-developed parallel anti-offset positioning algorithm, has been able to achieve nearly 100% detection accuracy, positioning and landing accuracy has been controlled within 0.25m, and (in the fastest case) can completed detection of 300 parking spaces within 20 minutes by a single UAV, the timeliness requirements are basically achieved.