As the development of autonomy in aerial robots, Micro Aerial Vehicle (MAV) has been more and more involved in our daily life. MAVs, especially quadrotors, have been widely used in field applications, such as disaster response, field surveillance, and search-and-rescue. For accomplishing such missions in challenging environments, the capability of navigating with full autonomy while avoiding unexpected obstacles is the most crucial requirement. In this thesis, we present methodologies, system designs, and cutting-edge applications, with a focus on motion planning, that enables a quadrotor autonomously navigate unknown complex indoor/outdoor environments using only onboard resources. We start by introducing algorithms for utilizing gradient information in the map to generate a sa...[ Read more ]

As the development of autonomy in aerial robots, Micro Aerial Vehicle (MAV) has been more and more involved in our daily life. MAVs, especially quadrotors, have been widely used in field applications, such as disaster response, field surveillance, and search-and-rescue. For accomplishing such missions in challenging environments, the capability of navigating with full autonomy while avoiding unexpected obstacles is the most crucial requirement. In this thesis, we present methodologies, system designs, and cutting-edge applications, with a focus on motion planning, that enables a quadrotor autonomously navigate unknown complex indoor/outdoor environments using only onboard resources. We start by introducing algorithms for utilizing gradient information in the map to generate a safe and smooth local trajectory. Then we investigate the problem of hard-constrained trajectory planning and propose an approach to generate MAV trajectory, which is guaranteed to be safe and kinodynamic feasible. Based on the above research, we further propose a planning framework which functions directly on point clouds, the most underlying data from different sensor type, without any post-processed maps. After that, we turn to study motion planning in the time domain and propose a method to generate the optimal time parametrization for spatial trajectories. Finally, we investigate and answer the problem of what is the best way to incorporate a human’s intention in autonomous and aggressive flight, and what is a flexible, robust and complete aerial system for the human involved applications. We present a system that integrates our past research on motion planning, and state-of-the-art perception and estimation, for aggressively flying in complex environments and capturing human’s intentions. Extensive experimental and benchmarked results, and detailed system set-up are presented throughout the thesis. We conclude this thesis by proposing future research opportunities.