Predictive control model allows drones to avoid obstacles and each other

There is a benefit to creating drones able to fly in swarms. If drones fly in swarms, they can cover larger areas and collect more data and information from any onboard sensors. Swarms would also allow drone designers to equip different units in the swarm with different types of sensors allowing them to gather a wider range of data.

However, a significant challenge in fielding swarms of drones is in preventing them from bumping into each other while allowing them to avoid obstacles. In a swarm of drones, when one drone has to change its trajectory to avoid an obstacle, neighboring drones synchronize their movements accordingly. However, this movement typically causes the swarm to slow down, creating a traffic jam of sorts within the swarm which can lead to collisions.

Researcher Enrica Soria has developed a new method to get around that problem in drone swarms. She developed a predictive control model that allows drones to react to others in the swarm while anticipating their own movements and predicting those of their neighbors. The researcher says the model gives the drones the ability to determine when a neighboring drone is about to slow down, so the slowdown has less of an impact on its own flight.

The model makes the drone less dependent on commands issued by a central computer. Rather than using a central computer, the model allows the drones to be commanded using local information and to modify their trajectories autonomously. Another interesting aspect of the control model is that tests show the system improves the speed, order, and safety of drone swarms in areas where there are lots of obstacles.

The technology has implications for several uses in the future, including drone light shows or search and rescue missions. Smaller drones can be less expensive than larger drones while swarms of smaller drones can cover more area making multiple smaller drone aircraft potentially more useful than larger single devices.