Amazon seeks human touch for delivery optimization
Amazon is teaming up with MIT to add driver intuition to the machine learning (ML) models used for optimizing delivery routes.
Amazon’s Last Mile team, which develops planning software for the e-tail giant’s delivery fleet, is s collaborating with the MIT Center for Transportation & Logistics (CTL) to incorporate driver know-how into route optimization models. The two groups are sponsoring a competition, called the Amazon Last Mile Routing Research Challenge, in which academic teams will train machine learning models to predict the delivery routes chosen by experienced drivers.
Amazon is providing the training data for the models and will be evaluating submissions, with technical support from MIT CTL scientists. The historical data provided by Amazon will include approximate delivery locations, package dimensions, and travel times and distances between locations, all information which is used by existing route optimization algorithms.
However, Amazon will also provide more than 4,000 traces of driver-determined routes, which encode the drivers’ know-how. Using both sources of information, contestants will be able to build models that identify and predict drivers’ deviations from routes computed in the traditional manner.
MIT CTL will publish and promote technical papers about the top-performing models. After the researchers have submitted their models, Amazon will release another 1,000 routes’ worth of historical data for evaluation purposes.
The winners’ prizes are $100,000 for first place, $50,000 for second, and $25,000 for third. Amazon may invite top-performing teams for research positions in the Last Mile organization, and top performers may also be invited to present their work at MIT CTL.
“We are encouraging participants to develop innovative approaches leveraging artificial intelligence, machine learning, deep learning, computer vision, and other non-conventional methods,” said Julian Pachon, director and chief scientist at Amazon Last Mile. “The contest is seeking to produce solutions to the route-sequencing problem that outperform traditional, optimization-driven operations research methods in terms of solution quality and computational cost.”