Coordinated Drone Systems for Exploration and Inspection

In this talk, I will begin by discussing our work on bounded rational drones that operate alongside other intelligent agents, such as other drones (or humans), in shared workspaces. To complete collective tasks, each drone must reason about the behaviors of its neighboring agents. However, real-world agents often do not act with perfect rationality due to various limitations, and predicting their optimal behavior can be both computationally demanding and error-prone. To overcome this, we utilize the concept of bounded rationality, enabling drones to account for and adapt to the suboptimal actions of others while staying within their own computational constraints. Next, I will present our recent progress in a hybrid system that integrates drones with mini ground robots. This system is designed for complex search and manipulation tasks, leveraging the complementary strengths of both aerial and ground-based robots. After the drone explores the environment and identifies hidden spaces with narrow openings, the mini ground robot can be deployed to navigate these confined areas for further exploration or manipulation. Finally, I will present a robotic information-gathering framework for environmental monitoring. This method enables the robot to use limited sensing data to efficiently characterize environmental attributes and construct 3D maps, such as terrain models, with broad applications in autonomous mapping, exploration, inspection, and navigation.