An architecture for mission coordination of heterogeneous robots?

Context: Robots can potentially collaborate to execute a variety of tasks in the service robots domain. However, developing applications of service robots can be complex due to the high level of uncertainty and required level of autonomy.<bold>Objective</bold>: We aim at contributing an architecture for the development of applications, capable of coordinating multi-robot missions, and that promotes modifiability and seamless integration of independently developed components.<bold>Method</bold>: In this work, we introduce MissionControl: an ensemble-based architecture to coordinate missions of heterogeneous robots to autonomously form coalitions. MissionControl comprises a component model and a runtime environment. The component model specifies how the system can be extended for different robot's behaviors and environments. The runtime environment provides the processes required for coordinating the execution of missions at runtime.<bold>Results</bold>: We evaluated MissionControl in a simulated environment in the healthcare domain. We randomly generated 81 scenarios with uncertainty in the robots' initial configurations. Then, each scenario was executed 8 times (i.e. 648 runs), where we evaluated the feasibility and efficiency of MissionControl for autonomously forming coalitions against a baseline approach that uses a random robot allocation. Statistical hypotheses testing yielded that MissionControl was able to achieve higher success rates while reducing the required time to conclude a mission, when compared to a baseline approach. We also perform an evaluation of the key quality attributes of the architecture, i.e. modifiability and integrability.<bold>Conclusions</bold>: MissionControl demonstrated itself able to coordinate multi-robot missions by autonomously assigning missions. Despite the error-prone robotic mission environment and demanding computational resources, MissionControl led to a significant increase in the success rate, while also decreasing the time required to conclude robotic missions when compared to a baseline approach. (c) 2022 Elsevier Inc. All rights reserved.