“Bees work in a colony,” Marc Kuo, founder and CEO of Routific, told Digital Trends. “They have scouts which go out to forage for nectar, which means exploring a lot of types of flowers over a long distance. As soon as a scout bee discovers a patch of flowers and returns to the hive with good quality nectar, it does a waggle dance to attract other bees which then go and explore that patch more closely.”
Inspired by this behavior, the algorithm Routific developed applies that concept to the world of logistics, by figuring out how a fleet of delivery vehicles can most optimally deliver packages to businesses or consumers. “The consumers are the flowers and the trucks are the bees,” Kuo continued. “What our algorithm figures out is the best route for the trucks to take to reach the consumers in a scenario where you might have 2,000 different addresses to deliver to and a fleet of 50 trucks. In that case, the question of how you sequence the delivery order to be optimal is incredibly complex.”
It’s essentially a variation on the long-studied “traveling salesman problem” — one of the best-known computer science algorithms, designed to figure out the optimal path between points. However, where the travelling salesman problem usually has only one salesman, in this case there are multiple different vehicles to consider.
“If you have just 57 addresses to deliver to, you already have more than a quattuorvigintillion possible route combinations,” Kuo continued. “That’s 1 with 75 zeroes after it. It’s impossible for humans to find the optimal route in that case, but even for an algorithm it’s next to impossible if you ask it to try every possible combination of routes in turn to figure out the best one. You need to be a bit more tactical about it — and that’s what bees have built into their nature, and we have built into our algorithm.”
Making things even tougher is the plethora of other challenges which need to be taken into account for deliveries, including whether a package needs to be delivered during a precise time window, the overall capacity of a truck, whether an item needs to be shipped in a refrigerated vehicle, and more.
Where the bee analogy comes into play is the way that the algorithm asks the computer to handle the searching task. “Our CPU is like a bee which has a bunch of areas it explores,” Kuo continued. “Whenever one area looks to be more promising, it gathers the attention of the other CPU power to direct more resources to that specific area to explore that specific search space a bit more. In that way we can find optimal routes, or routes that are very close to optimality, very, very quickly.”
It’s not just faster, either. Kuo also said that the routes his algorithm comes up with are typically 40-percent shorter than the manually planned routes many of his customers previously used. This has obvious positive impacts in terms of fuel savings, hours spent on the road, and the cost of vehicle maintenance. “In some cases we’ve even been able to take vehicles off the road because the original plan our customers have been working with have been so inefficient,” he said.
It’s no wonder Routific is creating a bit of a — dare we say it? — buzz.
Editors' Recommendations
- The future of cars: A new spin on an old idea could revolutionize autonomous vehicles
- How A.I. bumblebee brains could usher in a new era for navigation
- The BrambleBee robot promises to help honeybees pollinate flowers
- Man vs. machine: An A.I. algorithm attempts to break a world speed record
- 30 percent fewer taxis in New York City could be an improvement, MIT says
