Coordinated Remote Control of Unmanned Ground Vehicles (UGV) Using Autonomous Aircrafts (UAV)

For UAV-UGV collaborative systems, the UAVs must have the capability to observe the UGVs and control the UGV movements. If the UGVs are to be localised using the GNSS, then the localisation will be severely erroneous when the GNSS reception gets poor. Even if non-GNSS localisation techniques are used, pragmatic challenges arise if a UAV has to localise UGVs, e.g. localising UGVs operating under tree canopies. Therefore, the focus of this research is to develop methodologies to accurately localize and control the UGVs, which are suitable for harsh and GNSS-denied environments by collaboratively utilising UAVs. The developed novel methodologies can be used in applications such as autonomous forest fire fighting. The proposed UGV localisation method is a range sensing-based localisation method, where the range sensing is done from UAVs by ultra wide-band range sensors. Even though three UAVs are essential to localise a UGV which traverses on a horizontal plane, the proposed localisation method uses only two UAVs. Nevertheless, reducing the number of UAVs for UGV localisation makes the system deterministically unobservable. Therefore, a Constrained Extended Kalman Filter based localisation method and an Adaptive Extended Kalman Filter based localisation method are developed with novel kinematic constraints and heuristic constraints to overcome erroneous localisation due to unobservability. Using the same range sensing ultra wide-band links, the waypoints through which a UGV must traverse can be communicated to the UGV from UAVs, where a higher level path-planning will be executed in the UAVs to determine where the UGV must travel according to application requirements. Two novel robust Vector Field Navigation (VFN) methods are designed to use as guidance controllers to guide UGVs along the commanded waypoints, where the guidance algorithms are supposed to be executed in the UGV. The first VFN method designed guides a UGV such that the steering is induced by the vector field's flux interception rate, and the second VFN method designed guides a UGV based on optimal control principle such that the control reference trajectory is obtained from the vector field. A novel Fourier curve fitting-based vector field generation method is also developed to use the aforementioned novel VFN methods when the waypoints of the desired UGV path are to be received by a UGV in real time, i.e. when the UGV travel path cannot be predefined as a mathematical function