Automating Car Park Management With Blockchain And Computer Vision

Most current semi-automated parking lots are not able to cater for fully autonomous vehicles since traditional systems assume a human is driving the vehicle and making a payment physically. This empirical research aims to determine whether a system combining computer vision with blockchain technology, can be used to document parking occupancy in a transparent and efficient manner whilst facilitating payment autonomously. There are three main components to the proposed solution, which are: Identifying and Authorisation of vehicles; Occupancy Detection; and Payment handling. This study investigates the accuracy, scaling, and efficiency of all three components. Two parking lots, indoor/outdoor, and 5 different vehicles were used to simulate interactions that may occur whilst making use of a parking lot. Whilst noting that the identification and authorisation of vehicles is possible with an accuracy around 96%, the detection of occupancy and vacating of a parking space is also possible with an accuracy around 94%, we do not that the use of blockchain suffers from a volatile price index and limited transaction speeds. Analysis of the results showed that making use of a permission based blockchain, resulted in faster and cheaper transactions whilst sacrificing on decentralisation. The main components of the proposed solution are executed at the source, meaning a low-cost computer attached at each camera point is sufficient, resulting in a system which can scale upwards by design.