Introduction of the Flying Robots into the Human Environment: An Adaptive Square-Root Unscented Kalman Filter for a Fault Tolerant State Estimation in a Quadrotor

Recently, there have been growing interest in an autonomous control of Unmanned Aerial Vehicles (UAV). The key objective for scientists is to make the robots capable to operate in human shared environment. The main problem concerns control as well as filtering sensory data in all scenarios. Among many types of control algorithms there are a few which takes into account fault cases. The control algorithm can be fault tolerant in case of actuators saturation or their damage. The situation is more complicated in sensory system failure, in which case, the control is defenseless. In this article, a novel Adaptive Square-Root Unscented Kalman Filter (ASRUKF) in application of fault tolerant (FT) state estimator is presented. The SRUKF was designed for an orientation module of a quadrotor's mathematical model. The emphasis was put on the estimator's adaptability in case of the sensory system fault. The main objective of this paper was to show the idea of adaptability of the ASRUKF in terms of FT system. The work includes model derivation, explanation on the SRUKF algorithm as well as description of an adaptive parameters change of the estimator. Finally, the paper shows the experiment with a quadrotor in testbed and promising results.