We recently implemented a fairness constraint in our ODL Live vehicle route optimiser, for a passenger transportation routing problem. Getting an AI algorithm to schedule ideal routes for passenger transport often involves balancing multiple conflicting aims. In this case, we found that minimising the total of on-board time for all passengers (to maximise passenger comfort) led to unfairly long journeys for some passengers, although the majority had short journeys.

This can be solved in several different ways, each changing the objective function in the optimisation problem. We found the hard constraint illustrated in the below infographic to be effective for our case.

Generally though, there’s rarely a ‘one solution fits all’ for these types of problems, which is why we work closely with our clients, adapting ODL Live as needed to solve their specific routing problem.