|dc.description.abstract||The recent developments in research pertaining to the field of Unmanned Aerial Vehicles (UAVs) is motivated by its technical challenges as well as its practical implications in areas where human presence is inefficient, redundant or dangerous. The absence of human interference requires more robust and precise control techniques. However, most modern attitude control techniques require the knowledge of the current orientation of the body. There is no sensor available that explicitly measures the attitude
of a rigid body and hence, for small scale UAVs. it must be estimated using inertial vector measurements from low-cost and low-weight Micro-Electro-Mechanical System (MEMS) sensors like gyroscopes, accelerometers and magnetometers.
The predominant attitude representation formulations of a rigid body in three-dimensional space are recapitulated to elucidate the dynamical model of a quadrotor UAV. Low-cost MEMS are prone to significant noise effects from temperature change, vibrations, on-board magnetic fields generated by motors and currents. To improve the accuracy of the measurements sensor calibration techniques are explored. Primitive attitude estimation techniques like TRIAD, Davenports q-method, QUEST.FOAM, SVD method, etc. (which were aimed to be static optimization solutions to Wahbas Problem) were reviewed. These algorithms were extended to incorporate filtering techniques like Kahnan-type, to handle the measurement noise, and complementary filtering, where sensor measurements are fused
to reconstruct the orientation of a rigid body. Tlie latest nonlinear observers are also discussed for implementation purposes.
Practical implementation and performance comparison of various attitude estimation algorithms has been conducted on a small-scale quadrotor UAV, consisting of an inertial measurement unit (3-axis gyroscope, accelerometer and magnetometer), microcontroller, brushless motors, electronic speed controllers, on-board power supply and necessary frame constructs.||