Using drone swarms with manoeuvre units

We have examined the use of drone swarms in support of mechanized units at low tactical level. This has primarily been done through simulations. We have used the same interface for controlling the swarm in the simulator, as we use for controlling the swarm in the real world. The drones in our swarm are small drones of less than 10 kg, with a range of no more than 10 km. We have had up to 40 drones available simultaneously in the simulator. The purpose of our experiments have been twofold. On one hand, we want to see how a swarm can be used to assist a mechanized unit. This includes issues like what tasks the swarm could and should perform, and how a commander or battle captain best can use the information gathered by the swarm. On the other hand, we also want to improve our interface for controlling the swarm, and look at the role of the swarm operator (or operators). It is clear that a drone swarm, or even single drones, will help increase the Situational Awareness (SA) of the platoon or company commander. Having eyes in the sky enables us to look at areas behind terrain or other obstacles without exposing units to possible enemy direct fire. A swarm increases the SA beyond that of single drones, not only because there are more sensors available, but because a swarm operates in a smarter way than a group of individual drones. Screening an area with a swarm lets you detect enemies well before they get within line of sight. Sending a swarm out in front of the unit will let you know with decent certainty whether or not the area is safe to enter, and much faster than by scouting the area on foot. Since a swarm of small drones can cover a large area up to around 10 km away in a relatively short time, many possible targets may be detected. It is imperative that the drones have some sort of automatic detection / image recognition, as it is impossible for a drone operator to continuously watch the video stream from every drone in the swarm. Nevertheless, the operator is likely needed to verify or reject possible observations, both since hot spots in the terrain may register as a person or a vehicle, and because distinction between civilians and enemies so far is better done by an operator than a machine. In a conflict with many enemies, a swarm will detect many targets. This results in a need for means to destroy these targets, preferably without exposing own units. This can be done by various types of indirect fire, or by attack drones which may or may not be part of the swarm. In addition to detecting enemies, SA is also about knowing where the enemy is not. A drone swarm can cover a large area relatively fast, and what area has been covered is important data, even with no detections. It is still important to note that some targets may be too well hidden to be detected, especially in dense vegetation, so an understanding of the terrain is also important. In our experiments, few players have participated both on Blue and Red side. To better understand how a swarm best can be used in real settings and larger operations, experiments with more units, and thus more players, are needed. The simulations have provided valuable feedback on how to improve the interface for the swarm operator. During our work, the interface has been continuously changed and enhanced. Since the same interface is used for controlling drones in the real world, these changes have also been tested during live exercises. Using experiments to improve both the interface and the way the operator works, has proven very valuable.