Adaptive Swarm Operation uses a distributed algorithm using factor graph for deployment and coordination of UAV swarm in the environment. This algorithm supports the deployment of heterogeneous robots and facilities occasional arbitrary human input. Which means the same algorithm is used for the deployment of UAVs with different sensors, actuators and having a different operational regime with very little modification of codebase and changing parameters. Important behavioral features for the deployment operation has been encoded in the factor function of the factor graph implementation. For example in the LAPSE-RATE campaign important behaviors like collision avoidance, geo-fence boundary and following temperature gradients were used to fly coordinated flights of two Solo quadrotors equipped with radiosondes and a fixed-wing UAV all equipped with drop sondes. The navigation traces of some of the deployments have been shown in the plots.
Analysis of the data collected from OSU shows that adaptive operations are more efficient than manual vertical take of and landing. Adaptive operations have better information gain rate over time than manual deployments. In addition, it's smart behavioral design allows it to apply to the dynamic system of swarms. Moreover, it can offer human operators an intuitive interaction with the robots without interrupting the autonomy of the robots.