The ability of machines to learn from experience is the most valuable quality which allows robots to assist humans in every sphere and to solve extremely complicated tasks thus raising the quality of life. One of such tasks is easing traffic jams on toll highways in Malaysia. Nowadays it is being successfully delegated to artificial intelligence.
Kuala Lumpur, the capital of Malaysia, together with its agglomerations gives home to about 10 million people. The intense traffic has become a huge problem in the past years, taking into consideration numerous toll highways that let pass about 1.5 mln travellers per day.
The difficulty comes any time that drivers approach a toll plaza. There are several types of payment in Malaysia and each time the operator of the booth has to classify a car in order to apply the correct tariff. A car has to slow down for a few seconds while approaching the booth, thus creating a traffic jam.
The biggest toll highways operator in the country, the company PLUS, together with the leading AI developer NVIDIA have found an answer to the question. Now they are introducing a system called Triton which allows computers to identify the type, model and colour of a car (and read the licence plate) in just 50 milliseconds, which is about 10 times faster than a blink of a human’s eye.
The system is based on deep learning, that is a technology of machine learning imitating the way a human child gets certain types of knowledge. It allows a machine to form an abstract vision on the base of hierarchy of knowledge, received beforehand. Deep learning makes a computer less dependent on humans. The technology may be used in driverless cars, voice control devices, medical equipment (like smart beds in a hospital or cancer cell identifiers), aerospace industry, safety industry etc.