Abstract
This thesis investigates the application of heterogeneous robot teams in search and rescue (SAR) operations, aiming to enhance the efficiency and effectiveness of such missions. The overall objective is to explore the potential of heterogeneous human-robot teaming and identify ways in which robots can assist humans in performing SAR tasks. Firstly, the thesis explores the effective coordination of a heterogeneous robot team during search missions, where aerial and ground robots are deployed and assigned specific search tasks within a building. To optimize team performance, various mission goals are incorporated as constraints, and an assignment strategy based on Bipartite assignment is formulated. This approach leverages the navigation capabilities of different robots and ensures an efficient and targeted search operation. The proposed task assignment completely eliminates the chances for a human operator to be assigned to high-risk areas while maintaining a decent trade-off with a secondary mission objective. Secondly, the research focuses on enhancing the intelligent interaction between robots and human responders during SAR operations. This is achieved through the development of two essential motion planning algorithms for robots: a coordinated target-following algorithm and a robotguided evacuation algorithm. To accomplish this, the human-robot interactions are formulated using Model Predictive Control (MPC) and thoroughly validated through a combination of simulations and real-world experiments involving human testers. The proposed target-following algorithm effectively addresses 3D navigation constraints of its original algorithm and extends the original one-to-one target following to multi-to-one target following, allowing multiple robots to efficiently track and follow a single human target. The robot evacuation guidance algorithm combines both local path planning using MPC and a global path planner. This integrated approach successfully guides targets to safe exits even in complex simulated environments. The algorithm’s performance is also validated in real-world scenarios, where it outperforms the baseline in terms of how consistently the human evacuee could follow up with the guiding robot. These results suggest improvement in the efficiency and effectiveness of human-robot collaborative search and rescue emission, ultimately aiding human operators in saving more lives.