Mingqian Gao

From course material to a complete production workflow

In an AI project for vocational education, he contributed to domain-specific language-model training and a course-generation system, connecting material understanding, script generation, speech synthesis and digital-presenter video in one workflow. He also worked on model evaluation and inference optimisation. This experience in organising data, connecting modules and evaluating generated outputs informs his approach to mission-software design.

Adapting aerial grasping to different platforms

In NOEPRA’s aircraft-prototype project, he contributes to reproducing and adapting the Swooper high-speed aerial-grasping method, studying transfer across flight platforms and propulsion systems and working towards operation beyond motion-capture environments in unfamiliar indoor spaces.

The original Swooper paper reports grasping speeds of up to 1.5 m/s and an 84% success rate across 25 real-world trials. These are results from the original study, used to understand the method and reproduction conditions; the team records its own transfer results against its hardware, environment and trials.

He also contributes to a payload-adaptive flight-control design, studying control configurations as thrust, mass and centre of gravity change, and identifying the variables to compare when carrying different mission tools.

One interface for power, data and tool identity

He designed the circuit architecture for NOEPRA’s quick-change payload interface, considering power, communication, tool identification and status monitoring together, with connection requirements for visual sensing, thermal information and actuator control.

The interface design targets 6S flight platforms and multiple onboard communication channels. Lightweight tool designs target the few-hundred-gram range. Mass, power consumption and communication conditions are checked for each tool version, with integration trials used to assess their impact on the aircraft.

At a research camp at Shandong University, he led a team designing a vehicle-area safety electronic control system. The project combined thermal, proximity and ultrasonic sensors to study occupant detection and door-opening hazard warnings, and he contributed to the system design and demonstration-component testing. The prototype design was adopted by researchers associated with BYD. He subsequently documented the research in a Chinese-language article published in Auto Weekly, “On Vehicle Area Safety Electronic Control Systems” (《浅议车载区域安全电子控制系统》), discussing sensor and main-controller coordination, power-supply stability and environmental interference. These experiences provide a practical foundation for his work on onboard electronics, tool interfaces and status monitoring at NOEPRA.