SWARM RO-BAT-ics: building a robot analog of echolocating swarms

Try talking to someone in a loud concert crowd, you can barely hear yourself – forget what the other person is saying, right? In the concert setting, you can at least try again by repeating yourself or asking the other person to repeat themselves. Not all animals have that luxury, and the stakes can be higher. Echolocating bats for instance emit loud ultrasonic calls and listen to the returning echoes to detect their surroundings. When in groups, however, bats experience the sensory version of the rock concert setting – they can barely hear their own echoes over other bats’ loud calls. The stakes however are much higher, not detecting their own echoes, and thus objects around them can mean crashing into other bats (not so bad), or colliding into a cave wall for instance (drastic).

Though we know bats manage to fly in extremely large groups (hundreds to millions), how exactly they achieve this remains a largely unanswered question. Lab experiments are not possible at this scale, as it is impossible to house so many bats at one go, thus excluding the possibility of controlled experiments. Robots that mimic bat swarms are a neat alternative. We will build a ‘swarm Ro-BAT-ics’ platform to study how echolocating individuals are able to move despite being able to detect very little and only occasionally. Our pilot project aims to implement two working echolocating Ro-BATs to study (e.g., through synthetic evolution) which sensorimotor strategies are successful in large groups of active sensing individuals that experience sensory bottlenecks.