Collective Scenario Understanding in a Multivehicle System by Consensus Decision Making

In recent years, unmanned vehicles (UVs) have been largely employed in many applications. They, enhanced with computer vision and artificial intelligence, can autonomously recognize targets in an environment and detect events occurring in a real-world scenario. The employment of cooperative UVs can provide multiple interpretations supporting a multiperspective view of the scene. However, UV multiple interpretations often diverge, therefore, UVs need to find an agreed interpretation of the scenario. To this purpose, this paper proposes a novel consensus-based approach to lead multi-UV systems to find agreement on what they observe and build a group situation-based description of the scenario. UVs are modeled as experts in a group decision making problem that must decide on which situations best describe the scenario. First, the approach allows each UV to build high-level situations from the detected events through a fuzzy-based event aggregation. The event aggregation is modeled with a fuzzy ontology which allows each UV to express preferences on the situations. Then, a collective interpretation of situations is achieved by consensing each UV interpretation. Finally, consensus and proximity measures support the evaluation of the final group decision reliability. The assessed consensus reflects how much the collective scenario interpretation fits each UV perspective. The proximity measures support the detection of reliable and unreliable UVs to serve many tasks (i.e., mission replanning, damaged UV detection, etc.).