Session: 15-01-01: ASME International Undergraduate Research and Design Exposition

Paper Number: 150706

150706 - Tuning and Validating a Multi-Rotor Drone Model Using a Stationary Frame 

Accurately simulating multi-rotor drone flight is computationally difficult because of factors such as aerodynamic forces due to wind disturbances, weather conditions, and other factors related to controlling the drone. As such, there is interest in developing a high-fidelity multi-rotor drone model that can take into account many of these factors. A high-fidelity multi-rotor drone model has been previously developed to account for aerodynamic forces due to wind and the effects of different control systems. We want to improve upon this model by experimentally finding the relationship between motor thrust and rotor speed and using that relationship to tune the model. The tuned model will then be verified through real-world flight experiments. This work involves the following tasks: (1) designing and building a frame to collect stationary flight data, (2) conducting stationary flight experiments using the frame to obtain thrust and rotor speed data for tuning the model, and (3) conducting real-world flight experiments to validate the tuned model. To design the frame, we 3D-printed a custom clamping component that will make it possible to suspend the drone from the top of the frame, minimize horizontal movement, and connect a force sensor to the bottom of the drone to measure its thrust. We then built a frame within which the drone will be held stationary while performing different maneuvers. We measured motor outputs, power, rotor speed, and thrust as the drone mimicked movements it would do during a real flight trajectory. We used the data from these experiments to find the relationship between thrust, rotor speed, and power for our drone. This relationship was then used to tune the high-fidelity multi-rotor drone model. The results of the stationary experiments also informed the design of the real-world flight experiment trajectories since the relationships found gave insight into what types of trajectories had the largest effect on parameters such as power consumption. Then, real-world flight experiments were conducted to validate the tuned model. Specific trajectories were used to collect real-world flight data. These trajectories and the flight conditions were then replicated in simulation. The simulation results and real-world flight data were compared to validate the tuned model and demonstrate the differences between the tuned and untuned models. We expect the relationship between thrust and RPM from the stationary experiments to improve the simulation results compared to the untuned model. Additionally, the demonstrated process of tuning the model using a stationary frame can be applied to other drone models as a method for improving simulation results. The tuned model can be reliably used to predict drone responses to varying environmental conditions and control systems. The tuned model will also more accurately predict power consumption thus making it a useful tool when optimizing flight trajectories for efficient power consumption.

Presenting Author: Ian Agopsowicz Seattle University

Presenting Author Biography: My name is Ian Agopsowicz I was born and raised in Seattle, Washington. I am in my second year at Seattle University. I am majoring in mechanical engineering because I have always been interested in how everything works and everything automotive. I love racing on foot and in cars; I did four years of varsity running in high school and love going to the racetrack. I am also interested in pursuing research related to automotive engineering while at Seattle University.

Authors:

Ian Agopsowicz Seattle University
Samuel Mesfin Seattle University
Sameer Gill Seattle University
Marco-Antonio Sahagun Seattle University
Emile Edora Seattle University
Samantha Hoang Seattle University