Finding the fastest route between a number of locations is a well known problem for bees, as food sources are often spread out across a wide area around the beehive. And because a bee’s lifespan is measured in flight miles, they are highly incentivised to try to find the shortest route.
This means bees are constantly trying to solve what is known as the Travelling Salesperson Problem or TSP. The problem sounds quite simple, as all the bee needs to do is find the shortest route between a fixed number of locations, but its a problem which gets complicated very quickly. For example, if there are 5 locations, then there will be 60 possible routes, but if there are 10 locations, then there will be over 1.8 million options to choose from.
Scientists tested individual bees by moving artificial flowers to different locations and then monitoring how bees learned to fly between them. And it turns out bees were pretty good at gradually optimising their routes to shorten their flight times, and were certainly much better than humans, who typically have difficulty in both measuring distances as well as calculating and comparing routes based on the results.
But bees are better known for their collaborative problem solving. In a beehive the scouts look for food sources and then return to the hive and describe the food source and quality via a waggle dance. The onlooker bees then decide which scouts to follow, and the onlookers then become foragers as they go to fetch food.
Its a constantly evolving process, with onlookers voting for or against available food sources based on a constant flow of information from the scouts, and this allows the hive mind to efficiently optimise the gathering of food from multiple sources based on both distance and quality.
Bees use the same approach when deciding on a new location for the hive when they are about to swarm. Scouts will bring back news of multiple potential locations, but the hive voting system allows the hive to quickly decide on the optimal solution.
Mathematicians have tried to copy this technique and came up with the Artificial Bee Colony approach, which can be used to solve optimisation problems. Optimisation problems are problems where there are lots of possible solutions, and the complexity lies in trying to find the best one.
The Artificial Bee Colony approach, or ABC, involves trying to solve optimisation problems by simulating the multiple roles involved when bees forage for food. In the model simulation, each food source represents a possible solution to the problem and the quality of the food source represents the suitability of the solution. The individual ‘bees’ then evaluate the viability of solutions and their decisions then aggregate up to form the best solution.
An example of an optimisation problem is the travelling salesperson problem we started with. There are lots of different ways to approach this problem, but one option is to use the Artifical Bee Colony approach, where the overall solution is gradually optimised based on the multiple optimisations made by individual bees.
To be honest, its pretty complicated stuff, with several academic papers written on the subject, but its pretty impressive that we can use the hive mind approach of a beehive to try to solve complex real world problems.
Another approach to the Travelling Salesperson Problem is to use a technique called Self-Organising Maps, which also tries to hone in on the best solution. The gif below shows what the optimisation process looks like when using this approach to find the shortest route between 6,000 locations.
In this example the 6,000 locations are the red dots which make up Obama’s picture, and the blue line is a single continuous line which tries to fit all the points. In case you were wondering, the number of possible paths through 6,000 locations is a number with 20,000 digits and it is approximately 1 followed by 1498 zeroes.
