An academic approach to final mile headaches

Massachusetts Institute of Technology professors unveiled research conducted with the private sector to solve final mile delivery challenges in major cities at the university’s Center for Transportation & Logistics’ Crossroads event.    Two dynamics are conspiring to make final mile delivery increasingly fraught – cities are getting more crowded cities and demand for direct deliveries to consumers are rising.
   It’s making for a difficult situation for integrators and their retail and manufacturing customers. Sky-high service expectations, coupled with rock-bottom delivery price expectations, are leaving these supply chain partners in a bind. How will they get products into dense megacities in a timely manner?
   Much of the discussion at the MIT Center for Transportation & Logistics’ Crossroads symposium this week centered on how to meet these evolving supply chain challenges. The event, held annually in Cambridge, Mass., brings together academics, industry practitioners, and technology vendors to collectively think through long-term transportation and supply chain problems.
   The issue of final mile delivery in ever-larger cities was viewed through the lens of two Massachusetts Institute of Technology professors – Sertac Karaman, assistant professor of aeronautics and astronautics, and Matthias Winkenbach, director of MIT’s Megacity Logistics Lab.
   Karaman described the challenges in building and operating effective autonomous vehicles, noting, for instance, that the cost of a laser sensor, camera, and CPU-laden car could land you a nice Ferrari or two.
   And there are other issues with self-driving technology too – namely liability concerns in the event of an accident. Karaman said it’s unclear who would be considered responsible if an autonomous vehicle was at fault for a crash: the owner, the operator, or the manufacturer?
   Karaman said he sees development of the autonomous vehicle market mirroring the aircraft industry, where duplicative systems are built to ensure a failure of one component doesn’t cause a catastrophic incident.
   He also said advances in artificial intelligence are needed to support autonomous vehicles, noting that what separates existing autonomous vehicles from human driven ones is the ability to reason and process unusual variables.
   Karaman gave an example of how an autonomous vehicle can be programmed to stop if it sees a car or person in its field of view, but it might not be programmed to recognize a sofa in the middle of a highway.
   As part of an event exhibit, MIT displayed a tricycle that can be pedaled or driven by humans or run autonomously for multiple purposes in crowded cities. The cycle can be used for human conveyance when needed, similar to bike share systems across the world, but when not in use could be turned into package delivery vehicles.
   Meanwhile, Winkenbach explained that the problem of final mile delivery in megacities in emerging markets are the prevalence of so-called nanostores. These small single shops rarely carry more than a day or two of inventory, and so require almost daily (if not multiple daily) restocking efforts.
   Winkenbach’s research focused on multi-stop delivery patterns in Sao Paulo, Brazil, taking into account historical, social, and transactional data to paint a picture about how to structure the best sequence of deliveries. What may make sense from a miles-driven perspective may not work with when a nanostore actually needs stock. Or congestion in a certain part of a city at a certain time of day may necessitate using what would ostensibly be a less optimal delivery pattern.
   There are myriad physical factors at work like weather and elevation, but so many more intangible data points surrounding demand and micro-occurrences.
   What’s interesting about MIT’s work is that it is bridging the gap between the theoretical and operational. The university works closely with private sectors companies across a range of verticals, and so is developing research to underpin actual operational issues.