Fault-tolerant and collision-free maneuvering of unmanned aerial and ground vehicles for forest fire monitoring

Coordination and cooperation between unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) swarms has attracted significant research interest due to their complementary capabilities, enhanced operational effectiveness, and the broad range of applications enabled by such heterogeneous multi-agent systems. Building upon this complementarity and recent advances in consensus control for multi-agent systems (MAS), this paper seeks to design a fault-tolerant navigation strategy for a networked UGVs-UAVs system for tracking and monitoring a forest fire propagation. Firstly, a distributed and an adaptive formation tracking control is designed based on relative output measurements exchange among the agents namely the x and y coordinates. Thus, a Luenberger observer is introduced to estimate the neighboring relative error-states. Secondly, a new formation model design is proposed using the dynamic Elliptical Fourier Descriptors (EFDs) yielding high flexibility and adaptability when tracking the fire propagation contour. The communication topology among UGVs-UAVs agents is assumed to be dynamic, directed and must simply contain a directed spanning tree, which is a weak requirement in real applications. In case a fault occurs to one or a set of agents in the team, a formation reconfiguration approach is incorporated based on a task reassignment algorithm, resulting in a robust formation tracking during the fire tracking mission. Additionally, a new collision avoidance method is proposed for the UGVs based on the parametric and implicit features of the locally sensed obstacles. Finally, Lyapunov method is used to analyze the stability of the overall UGVs-UAVs system. Two simulation examples illustrate the feasibility and the effectiveness of the proposed navigation framework under switching communication topologies, where the formation vehicles are supposed to have different communication capabilities reflecting real-world operating conditions.