Sensor Fault Detection and Estimation for Quadrotors Using Kinematic Equations

Abstract

Obstacle avoidance of Micro Aerial Vehicle (MAV) in urban environment is the most complex, difficult and essential part of the autonomous flight problems. The paper presents a simple ad-hoc strategy using a pair of miniature laser rangefinders (i.e. MLR100) and two PIDs cooperating with an obstacle avoidance controller. The strategy can be realized as an additional routine integrated with the autopilots firmware (i.e. MP2128HELI). The main advantage of the proposed strategy is simplicity of its implementation in small-sized MAVs and its power efficiency. All previous works, especially the vision-based ones require high performance microprocessors which is an important limitation when applying on real MAVs. On the other hand, the autonomous controller, which is based on optic flow sensors, is easy to implement on even tiny MAVs, but optic flow sensors require applicable level of contrast variation, so their performance is strongly sensitive to weather conditions. The proposed idea of the autonomous obstacle avoidance system in urban environment was simulated using MATLAB SIMULINK software. In the real flight all computations and controls will be realized by the advanced autopilot, hence the rest of autonomous control and complex flight dynamics are not included in the simulation. The assumption allows to spot a more focused attention on the obstacle avoidance problem and a simpler model of the MAV can be used in the simulation. The results presenting the 2D trajectories confirm that the effectiveness and safety of the proposed strategy of obstacle avoidance is attainable during the real flight in streets canyons.