Date of Award
Thesis - Open Access
Master of Science in Aerospace Engineering
Dr. Richard Prazenica
First Committee Member
Dr. Hever Moncayo
Second Committee Member
Dr. Troy Henderson
Electro-optical devices have received considerable interest due to their light weight, low cost, and low algorithm requirements with respect to computational power. In this thesis, vision-based guidance laws are developed to provide sense and avoid capabilities for unmanned aerial vehicles (UAVs) operating in complex environments with multiple static and dynamic collision threats. These collision avoidance guidance laws are based on the principle of proportional navigation (Pro-Nav), which states that a UAV is on a collision course with another vehicle or object if the line-of-sight (LOS) angles to the object remain constant. The guidance laws are designed for use with monocular electro-optical devices, which provide information on the LOS angles to potential collision threats, but not the range. The development of these guidance laws propagates from an investigation into numerous methods of Pro-Nav based guidance, including the use of LOS rate thresholding, avoidance of the most imminent threat detected, and objective-based cost optimization. The collision avoidance guidance laws were applied to nonlinear, six degree-of-freedom UAV models in various simulation environments including a varying number of static and dynamic obstacles. A final form of the avoidance law, determined from these simulation studies, was applied to a small-scale UAV model flying through a virtual urban environment, which utilizes camera-in-the-loop simulation techniques.
The final results of these studies showed that the most effective approach was to implement a cost function-based avoidance law that includes a term based on the Pro-Nav intercept heading for a desired waypoint and avoidance terms for all obstacles in view that pose a collision threat. Obstacle avoidance headings in the cost function are based on the difference in the obstacle LOS rates from the magnitude of the minimum safe LOS rate. When applied to UAV simulations in a virtual urban environment, this guidance law provided successful avoidance for the case of a single building, maintained a safe heading through an urban canyon, and determined the safest path through a complex urban layout. For the case of the complex urban layout, a single collision during flight occurred due to a lack of visual feature points to contribute to the avoidance law calculation.
Scholarly Commons Citation
Clark, Matthew J., "Collision Avoidance and Navigation of UAS Using Vision-Based Proportional Navigation" (2017). PhD Dissertations and Master's Theses. 323.