From drones that transport blood samples to cube robots that dispatch groceries, automated and autonomous small vehicles are streamlining delivery services. In efforts to further optimize vehicle-based transport, sidekick routing has emerged as a logistical framework in which a base vehicle such as a truck or van hosts several ‘sidekicks’ – smaller ‘helpers’ like unmanned aerial vehicles (UAVs) or autonomous ground vehicles (AGVs). These sidekicks move between picking up items, visiting customers, and visiting the base vehicle. As demonstrated by companies like Kiwi Campus, Dispatch, and even government actors like the Swiss Post, this modular system is becoming increasingly prevalent in both the private and public spheres for its potential to reduce vehicle miles traveled (VMT) and increase access to the last mile. The figure below illustrates the potential for route optimization. Route A’s ‘traveling salesman tour’ features a single truck visiting a set of points from a central location and is almost double the length of route B, in which the truck utilizes a sidekick. Vehicles in a ‘traveling salesman tour’ not only sacrifice time and resources, but also face significant limitations regarding accessibility. Sidekicks can travel down narrow alleyways and corridors that are simply inaccessible to a full-sized delivery truck.

While the physical hardware for autonomous delivery is at an advanced stage and the addition of a sidekick clearly provides notable improvements, coordinating a set of sidekicks poses its own difficulties. Sidekicks travel simultaneously at differentiated speeds to multiple locations. However, to get the most out of any automated system, simplifying operation and optimizing its adaptability is essential. When it comes to the successful implementation of autonomous delivery services, the principal obstacle is the route itself. How can you determine the best route – one which makes maximal use of the sidekicks, minimizes mileage, and increases speed? With funding provided by the U.S. Department of Transportation, PSR researchers John Gunnar Carlsson, Kellner Family Associate Professor at the University of Southern California in the Epstein Department of Industrial and Systems Engineering, and USC graduate student Bo Jones, set out to answer precisely this sort of questioning a research project, titled “The ‘Sidekick’ Routing Paradigm for VMT Reduction and Improved Accessibility.”

Carlsson and Jones conducted their research in three phases. First, they developed a mathematical model for calculating efficient routes. Second, they replaced difficult-to-compute formulas with simplified formulas that provide accurate estimations under a set of specific circumstances. In this case, they also assumed fixed vehicle and sidekick speeds, single trips to and from the base vehicle, unlimited drone range, and the possibility of slower drone speeds in comparison to a truck or van. This allowed the researchers to focus on a manageable number of variables, which in turn allowed them to conduct computational experiments to obtain their results and subsequently produce a formula to guide operators in route development.

Considering the importance of speed and number of sidekicks, Carlsson and Jones sought to find ways to better predict the outcomes of different scenarios including simultaneous truck and sidekick delivery, any potential number of sidekicks, and any location point.

Through their research, Carlsson and Jones have made a significant contribution to the broader literature. Their project stands out among myriad efforts to both optimize the efficiency of autonomous delivery systems and ensure that more customers have access to such services. Flexible and widely applicable, Carlsson and Jones’s routing paradigm is ripe for adoption by both public and private sectors.