Learn how your company can create applications to automate tasks and generate further efficiencies through low-code/no-code tools on November 9 at the virtual Low-Code/No-Code Summit. Register here.
Researchers from Stanford University have devised a way for hundreds of drones to use the bus or trams in an effort to redesign how packages are distributed in cities. Should such a solution ever scale, it could reduce delivery van congestion and energy usage while extending the distance a drone can travel to deliver a package.
There’s a reason most delivery drones we’ve seen thus far are dropping packages off in the suburbs. Urban centers can be dynamic environments, full of unexpected obstacles, and drones are still not permitted to fly freely through cities. But researchers say using public transportation can increase a drone’s range up to 360% beyond travel with flight alone.
“Our approach strives to minimize the maximum time to complete any delivery,” the team writes in a paper published this week at the online 2020 IEEE International Conference on Robotics and Automation (ICRA). “By combining the strengths of both, we can achieve significant commercial benefits and social impact.”
This approach, which involves the drones hitching a ride on the outside of buses and trams, could help overcome the limited travel capacity of drones today. The popular DJI Mavic 2, for example, is able to fly a maximum distance of 18 kilometers, or about 11 miles round trip.
Join today’s leading executives at the Low-Code/No-Code Summit virtually on November 9. Register for your free pass today.
The Stanford system could handle up to 200 drones delivering up to 5,000 packages. The AI network is made for cities with up to 8,000 stops, and experiments were conducted specifically in San Francisco and Washington, D.C. For context, the San Francisco Municipal Transportation Agency (SFMTA) covers an area of 150km2 and the Washington Metropolitan Area Transit Authority (WMATA) covers an area of roughly 400km2.
The multi-drone network does not include use of SFMTA or WMATA tunnels. Paper coauthor Shushman Choudhury told VentureBeat in an email that simulations do not take into account any physical infrastructure, and instead relied on open source data on bus stops and drone package depot locations. Researchers did not consult SFMTA or WMATA officials, but that could make sense after further research to discover additional externalities or potential impact on urban communities.
The authors describe the solution as resembling algorithms on-demand mobility services developed to coordinate multiple modes of transportation. Like Uber, Lyft, or other companies that combine ride-sharing options with public transportation, electric scooters, and walking, the model takes a two-layered approach.
“First, the upper layer assigns drones to package delivery sequences with a near-optimal polynomial-time task allocation algorithm. Then the lower layer executes the allocation by periodically routing the fleet over the transit network while employing efficient bounded-suboptimal multi-agent pathfinding techniques tailored to our setting,” the paper reads.
The research comes out of the Stanford Intelligent Systems Laboratory (SISL) and Autonomous Systems Lab. Titled “Efficient Large-Scale Multi-Drone Delivery using Transit Networks,” the work was nominated by ICRA conference organizers for best multi-robot systems paper. Authors of the paper, including Choudhury and Mykel Kochenderfer, published research last year about an AI technique called DREAMR that’s capable of guiding a single drone, using buses and trams to reduce flight time and conserve energy.
The multi-drone approach detailed at ICRA this week assumes packages can be acquired from any dispatch depot. It also assumes a drone will carry one package at a time and that drones will recharge or replace batteries at depots when time allows. Next steps could include factoring in issues like delays and ideal travel time windows. Anybody who’s ridden a Muni bus in San Francisco knows traffic and congestion can considerably chew into travel times.
“A key future direction is to perform case studies that estimate the operational cost of our framework, evaluate its impact on road congestion, and consider potential externalities, like noise pollution and disparate impact on urban communities,” the paper reads.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.