Analysis of ground infrastructure and sensing strategies for all-weather approach and landing in Urban Air Mobility

As a key requirement for the development of a safe Urban Air Mobility (UAM) framework, all-time/all-weather navigation and perception capabilities must be ensured to Vertical Take-Off and Landing (VTOL) aircraft, especially in the approach and landing phases. In these scenarios, the conventional landing systems are expected to be complemented or replaced by technological solutions tailored to UAM, including the use of multiple exteroceptive sensors and, consequently, of multi-sensor navigation algorithms. Furthermore, fully autonomous landing capabilities represent a prerequisite to gradually remove the necessity for onboard pilots while maintaining safe operations. Within this framework, after a brief overview of the possible navigation solutions both in terms of required onboard sensors and ground infrastructure, this paper focuses on the integration of Artificial Intelligence (AI) techniques in a vision-based navigation architecture. A Deep Learning based object detector is trained to recognize conventional landing pads and integrated within a pose determination pipeline to generate aiding measurements for a multi-sensor state estimation filter. The performance of the implemented architecture is assessed in simulated scenarios. Finally, initial research efforts relevant to the integration of Frequency Modulated Continuous Wave (FMCW) radars within the onboard sensing suite are presented, which deal with the generation of simulated radar data in customizable urban scenarios.