Single chip solution for stabilization control & monocular visual servoing of small-scale quadrotor helicopter

Hossain, Mohammed Raju (2017) Single chip solution for stabilization control & monocular visual servoing of small-scale quadrotor helicopter. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (72MB)

Abstract

This thesis documents the research undertaken to develop a high-performing design of a small-scale quadrotor (four-rotor) helicopter capable of delivering the speed and robustness required for agile motion while also featuring an autonomous visual servoing capability within the size, weight, and power (SWaP) constraint package. The state of the art research was reviewed, and the areas in the existing design methodologies that can potentially be improved were identified, which included development of a comprehensive dynamics model of quadrotor, design and construction of a performance optimized prototype vehicle, high-performance actuator design, design of a robust attitude stabilization controller, and a single chip solution for autonomous vision based position control. The gaps in the current art of designing each component were addressed individually. The outcomes of the corresponding development activities include a high-fidelity dynamics and control model of the vehicle. The model was developed using multi-body bond graph modeling approach to incorporate the dynamic interactions between the frame body and propulsion system. Using an algorithmic size, payload capacity, and flight endurance optimization approach, a quadrotor prototype was designed and constructed. In order to conform to the optimized geometric and performance parameters, the frame of the prototype was constructed using printed circuit board (PCB) technology and processing power was integrated using a single chip field programmable gate array (FPGA) technology. Furthermore, to actuate the quadrotor at a high update rate while also improving the power efficiency of the actuation system, a ground up FPGA based brushless direct current (BLDC) motor driver was designed using a low-loss commutation scheme and hall effect sensors. A proportional-integral-derivative (PID) technology based closed loop motor speed controller was also implemented in the same FPGA hardware for precise speed control of the motors. In addition, a novel control law was formulated for robust attitude stabilization by adopting a cascaded architecture of active disturbance rejection control (ADRC) technology and PID control technology. Using the same single FPGA chip to drive an on-board downward looking camera, a monocular visual servoing solution was developed to integrate an autonomous position control feature with the quadrotor. Accordingly, a numerically simple relative position estimation technique was implemented in FPGA hardware that relies on a passive landmark/target for 3-D position estimation. The functionality and effectiveness of the synthesized design were evaluated by performance benchmarking experiments conducted on each individual component as well as on the complete system constructed from these components. It was observed that the proposed small-scale quadrotor, even though just 43 cm in diameter, can lift 434 gm of payload while operating for 18 min. Among the ground up designed components, the FPGA based motor driver demonstrated a maximum of 4% improvement in the power consumption and at the same time can handle a command update at a rate of 16 kHz. The cascaded attitude stabilization controller can asymptotically stabilize the vehicle within 426 ms of the command update. Robust control performance under stochastic wind gusts is also observed from the stabilization controller. Finally, the single chip FPGA based monocular visual servoing solution can estimate pose information at the camera rate of 37 fps and accordingly the quadrotor can autonomously climb/descend and/or hover over a passive target.

Actions (login required)

View Item