State Estimation and Multi-Sensor Data Fusion for Micro Aerial Vehicles Navigation

Data fusion is the process of combining measurements from multiple sensors into a more accurate and reliable estimation of the desired state. We exploit the role of IMU sensors in the autonomous navigation of a micro quadrotor and its integration with exteroceptive sensors such as small laser range-finders and low cost RGB-D (color plus depth) cameras for real time pose estimation.;An IMU is a combination of a tri-axis gyroscope, which measures angular velocity, and a tri-axis accelerometer, which measures linear acceleration. Additionally, if the IMU has a tri-axis magnetometer, it is referred to as MARG (Magnetic Angular Rate and Gravity). An inertial navigation system (INS) adopts IMUs to recover position, velocity and orientation from integration of acceleration and angular velocity readings. However, because of low-varying biases affecting the measurements, integration causes accumulation of error over time making the estimation inaccurate. For this reason, the combination of IMU measurements with data from external sensors is necessary to improve the quality of the estimation. MAVs present a set of fundamental challenges such as limited payload and computational power. Therefore, the sensor fusion algorithm needs to be computationally efficient to run onboard the vehicle and able to reduce noise and drift that typically affect measurements from low cost sensors.;In this work, we present methods to combine data from different sources with focus on orientation estimation algorithms. We first propose a complete navigation system for indoors navigation of a microquadrotor using a combination of data from the onboard IMU and a laser range-finder. We address the issues of autonomous control, state estimation, path-planning and teleoperation. We then test the same system using different exteroceptive sensors such as a RGB-D and a recently developed mobile device from Google's Tango project.;We present the algebraic quaternion algorithm (AQUA) - a novel mathematical approach to estimate the orientation of a rigid body in quaternion form from gravity and magnetic field observations. The simple algebraic derivation of the quaternion allows us do develop novel, efficient data fusion algorithms suitable for MAV orientation estimation fusing acceleration and magnetic field with angular velocity data. We validate the novel orientation estimation approach during an indoor MAV flight experiment using ground truth data from a motion capture system. We compared our results against other algorithms, showing better performances of our method.