Fei Gao

Fei Gao's research focuses on the challenges of autonomous flight for UAVs in complex and dynamic environments, and he has achieved a series of systematic results centered on the two core pillars of motion planning and perception-planning closed-loop.

Facing the challenges of high computational load and poor real-time performance in traditional motion planning, Fei Gao proposed a spatio-temporal optimal trajectory generation framework. This method efficiently generates smooth, dynamically feasible trajectories, improving planning efficiency by several orders of magnitude. He has also introduced reinforcement learning, enabling UAVs to adaptively adjust their flight strategies for high-speed autonomous flight.

To address the challenge of UAVs flying with only onboard sensors and limited computation, he took a perception-planning closed-loop perspective. He and his team developed an event camera inspired by saccadic eye movements to enhance high-dynamic perception. Furthermore, he proposed a local obstacle avoidance framework based on sparse collision checking that eliminates the need to build dense environment maps, thereby addressing industry challenges related to high-dynamic perception and the cost of environment modeling.

His research has also led to significant breakthroughs in drone swarm collaboration. In 2022, he led his team in achieving the autonomous flight and collaborative exploration of a decentralized drone swarm in an unknown wild environment. Recently, he has expanded his research into flying embodied intelligence by proposing a general decision-making framework for flying robots in complex environments and versatile tasks. His work provides core algorithmic support for flying robots, promoting their practical application in scenarios such as aerial inspection, search and rescue, and logistics.