Swarms of small flying robots have great potential for improving disaster research and rescue missions. Their tiny size means that they can manoeuvre in small spaces. Swarms of drones could enter a building, explore it, and come back to the base station with relevant information.
Up to now, the lack of adequate navigation strategies has put this task out of reach. But after four years of work, researchers from Delft University of Technology (of the MAV Lab), Radboud University and Liverpool University are presenting a solution called the ‘Swarm Insect Algorithm’ that allows a swarm of small flying robots to autonomously explore an unfamiliar environment. Their work was published in Science Robotics on 23 October.
As part of the project, researchers equipped small 33 gram commercial ‘off-the-shelf’ drones with cameras and deployed them in an indoor office environment to find two dummies that represent victims in a disaster scenario. Within six minutes, a swarm of six drones could explore about 80% of the open spaces - which would be impossible for one drone alone.
The team designed a method for the small drones to detect and avoid each other by having each drone carry a wireless communication chip and then using the signal strength between these chips.
‘Bug algorithm’ inspired by insect navigation
The biggest challenge was to let the small robots navigate in an unfamiliar environment by themselves. The reason for this is that small robots have very limited sensing and computation capacity. The team tackled this by developing their new ‘bug algorithm’ inspired by the relative simplicity of insect navigation.
Insects do not make highly detailed maps of the environment. Instead, they deal with obstacles on the fly. In principle, detailed maps are very convenient, because they allow a robot to navigate from any point on the map to any other point along an optimal path. However, the costs of making maps on tiny robots is prohibitive. The proposed bug algorithm may lead to more circuitous routing, but its merit is that it can be run even on tiny robots.
The project was funded by the Dutch national scientific foundation’s (NWO) Natural Artificial Intelligence programme.
McGuire et al., Minimal navigation solution for a swarm of tiny flying robots to explore an unknown environment, Science Robotics, 2019