Optimal Waypoint Guidance for Collision Scenarios

open_access
Open Access
Author:
Perumalla, Aniruddha
Graduate Program:
Aerospace Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 02, 2023
Committee Members:
Amy Pritchett, Program Head/Chair
Eric Norman Johnson, Thesis Advisor/Co-Advisor
Jacob Willem Langelaan, Committee Member
Keywords:
UAV
collision avoidance
autonomy
aircraft
aviation
Abstract:
As unmanned aerial vehicles (UAVs) grow in popularity for industrial, military, and civilian applications, the need for such aircraft to possess safe and reliable collision avoidance system rises in importance. This thesis proposes a collision avoidance system for UAVs to avoid unexpected collision with static and dynamic obstacles. The proposed system can adhere to airspace regulations specifying safety zones around other aircraft, is simple enough for deployment in a "plug-and-play" fashion onboard a UAV, and is efficient in terms of control effort needed to achieve the avoidance. The system contains logic to determine if a given obstacle is threatening, as well as to generate an "aiming point" to which the vehicle should travel to avoid the obstacle; it accomplishes this by examining the geometry of the encounter and transposing the problem of collision avoidance to one of waypoint guidance. Furthermore, the system contains a simple feedback controller that can guide the ownship to such an aiming point or a generic goal in the flight environment, while minimizing the overall control effort (e.g., fuel) needed to do so. High-fidelity simulation was used to validate the performance of the collision avoidance framework. The potential of the framework to enable UAVs to avoid head-on collision, even at high speeds, was also successfully demonstrated through flight tests at a university research site at an airport. The framework developed in this thesis has been implemented in working software and validated in simulation. Furthermore, it has been implemented and deployed in the onboard controller of an experimental helicopter UAV. In a series of flight tests at an airport site, the proposed framework controlled the helicopter to successfully avoid head-on collision with a "ghost" dynamic intruder UAV at relatively close distance and moving at high speed. In the flight tests, the helicopter and ghost intruder were initially separated by 1 mile at the outset of the encounter and flew at relative speeds in excess of 100 feet per second. Because this framework is able to be integrated with other flight controllers in a simple fashion, it relieves flight controllers tailored for long-term strategy from the short-term tactical burden of unexpected collision avoidance challenges. This also means long-term planning may be greatly simplified in execution by primarily focusing on global cost optimization, enabling longer missions and/or reduced demands on reserves.