Humanoid-Coordinated Vertiport Terminal Operations

 (a) Problem Statement: An unparalleled lack of ground capacity exists in global megacities, while air capacity solutions, like helicopters, are not sustainable due to being too expensive and noisy. Current automatic solutions have an ability to coordinate only small-size swarms without having social awareness and quick enough sensor fusion for dense corridors.

(b) Solution: In this paper, the authors suggest a symbiotic solution relying on zero-emissions autonomous eVTOL vehicle fleets and socially-aware humanoid coordinators on the ground. Coordination of aircrafts is performed using decentralized actor-critic multi-agent reinforcement learning network. Humanoid agents are taught to navigate through crowds and help passengers using NVIDIA Omniverse digital twins.

(c) Significant Findings: According to the hypothesis, the implementation of the decentralized MARL scheduling algorithm may help to cut down approach-corridor conflicts by more than 40%. Moreover, the use of humanoid coordinators, operating together with central airport management, allows increasing terminal capacity up to 80-120 flights/hour through the use of IoT edge-nodes that provide sensor data fusion within 50 milliseconds.

(d) Applications: This approach opens the way for future urban air mobility infrastructure providing human-centric solutions for security verification, cargo management, and passenger assistance in tight rooftop