Improving TRISAT-R Satellite Attitude Prediction by Incorporating a High-Resolution Optical Sensor to Kalman Filtering

When flying a satellite, various operations on-board the satellite require determining quite a precise orientation to properly function, such as Earth or Sun observation, downlink communication and similar. This results in satellite attitude estimation posing a crucial problem to be solved, using a wide array of spacecraft control systems and sensor readings. Due to the noise in magnetic sensor measurements and drift in gyro meters, auxiliary (optical) subsystems are widely used to add reference measurements and correct the noise and the drift. Since the star-trackers, most widely used optical subsystem for attitude estimation, can be relatively large and high energy consuming, a smaller alternative is found in photodiodes and small CMOS cameras, both of which are being currently used on board TRISAT-R. Even with additional optical filters, there is error present in all measurements and error estimation is necessary. Recently the development and use of Unscented Kalman Filter (UKF) has shown greater potential for use in space related estimation, due to UKF not using classic Kalman linearization but employing a set of sigma points to calculate the systems mean and covariance more accurately. TRISAT-R use of on-board systems to tackle the problem of attitude estimation is described. UKF filtering is briefly discussed, and the simulation model for UKF of TRISAT-R orbiting is proposed. The results show an increase in accuracy of attitude prediction when using CMOS cameras for prediction instead of photodiodes, which confirms the idea that small cameras can be used for in-orbit photodiode calibration.