A Vision-Based Approach to Autonomous Landing of an eVTOL Aircraft in GPS-Denied Environments

This research paper describes an autonomous system developed by a team of undergraduate engineering students at Embry-Riddle Aeronautical University - Prescott with support from Wisk Aero. Autonomous systems enable an electric vertical takeoff and landing (eVTOL) aircraft to perform the fundamental duties of a pilot - aviate, navigate, and communicate - without human intervention. A prototype eVTOL system, implemented as a modular payload attached to the bottom of a commercially available drone, was developed and tested to operate from the final approach through the landing phase of flight. The prototype vision-based system identifies a landing zone, and calculates relative position and orientation position without reliance on Global Positioning System (GPS) capabilities, which is potentially vulnerable to interruption and disruption. In addition, the system utilizes a machine learning approach to identify if any obstructions impact the safety of landing the aircraft. Through the development of the prototype, the findings demonstrated the viability of a computer vision-based landing system to be used on an eVTOL aircraft for the purpose of autonomous advanced air mobility (AAM) solutions.