Swarm control for collaborative continuous searching with “autonomous and adaptive control”

Autonomous swarm control, which is expected to be able to handle multiple unmanned vehicles (UxVs), is an important technology that can be applied to many operations and incorporated in many systems. For operations in real environments, we have to consider severe conditions under which communications between UxVs may be inadequate and the environment changes unexpectedly. In addition to optimality, adaptability (meaning robustness to severe and unexpected conditions) is important in real situations. However, there are no quantitative benchmarks for algorithms to have both optimality and adaptability. Here, to reconcile optimality and adaptability, we have developed an “autonomous and adaptive control”. The distinctive features of this algorithm is that it simply uses the purposes of a system, such as efficiencies, to control elements and it has a distributed optimization architecture without a centralized control unit. We think that it is promising for autonomous swarm control with optimality and adaptability in real environments. In this paper, we benchmark several swarm control algorithms including our own. We compare two conventional algorithms with ours in a scenario involving a continuous search operation where the UxVs keep moving and searching. We also devise performance indicators for this operation and evaluate the algorithms in terms of them. The results obtained from computational simulations show that our algorithm maintains the efficiency of the search operation at a high level. That means the algorithm is a promising approach for achieving both optimality and adaptability.