Hong Kong’s metro system is using artificial intelligence to coordinate the movements of its subway engineers with optimised efficiency and speed.
Performing maintenance on Hong Kong’s sprawling subway system presents an immense challenge, particularly given the extensiveness of the city’s underground railway network and the sheer density of the urban population which it serves.
Despite the complexity and scale of the task at hand, owner MTR Corporation runs an operation of exceptional efficiency which is renowned as one of the world’s most punctual subway systems, with a 99.9 per cent on time rate.
Achieving this level of efficiency requires a staff 10,000 workers who perform as many as 2,600 engineering fixes a week, largely during nocturnal shifts which only commence once the last subway train completes its final journey just after midnight.
In order to coordinate this army of underground metro engineers, MTR Corporation has made recourse to an artificial intelligence system which is capable of scheduling their work tasks with far greater speed and efficiency than the human mind can achieve.
Andy Chan from Hong Kong’s City University designed the system in collaboration with MTR, and was responsible for incorporating it into the planning and coordination of its maintenance schedule.
According to Chan, prior to the use of the AI system, maintenance works were haphazard affairs, entailing planning sessions with experts from five or six different areas.
His program uses a comprehensive simulation of the underground rail network to determine the optimum schedule for performing required engineering work – a feat which can be achieved due to the ability of the AI software to view the entire system from an omniscient perspective, as opposed to the limited viewpoints of individual human experts.
The AI system employs what is referred to as a genetic algorithm to find the optimal planning solution, The algorithm devises multiple solutions to the same problem before pitting them against each other in a “survival of the fittest” competition to determine which is the best.
The AI program has achieved considerable savings for MTR, reducing the time needed to plan the repair schedule by two days per week and providing repair teams with an additional 30 minutes to complete each of their nocturnal shifts.
Despite superseding human beings in its ability to schedule engineering work, the system was originally dependent upon human expertise to achieve its peerless levels of efficiency. Chan interrogated experts about the factors they weigh and consider when making decisions with respect to engineering maintenance work before incorporating this data into the constituent algorithms of the AI program.
Given the tremendous success of the AI scheduling program in Hong Kong, MTR Corporation is now planning to introduce the software to the other networks it manages around the world, which include subways systems in Beijing, London and Melbourne and Stockholm.