MULTI-SENSORY MOTION ESTIMATION AND CONTROL OF A MINI-QUADROTOR IN AN AIR-

GROUND MULTI-ROBOT SYSTEM

Tianguang Zhang, Wei Li, Markus Achtelik, Kolja K ¨uhnlenz and Martin Buss

Unmanned Aerial Vehicles (UAVs) are a major focus of active researches, since they can extend our capability in a variety of areas. A significant challenge in developing UAVs is to extract and fuse the useful information in a robust manner in order to achieve a stable flight and an accurate navigation. On-board Inertial Measurement Units (IMUs) are commonly equipped. However, the drift of inertial sensors leads to errors during integration over time, making a steadily accurate estimation of the absolute pose nearly impossible. Therefore, most works rely on an external infrastructures, e.g. GPS for outdoor applications [3] [4] and tracking system for indoor environments [5] [6]. Vision and proprioception are the primary senses to guide the movement. Taking insects as an example, the both senses are important to be fused for localization and motion estimation. They are able to complement the limitations and deficiencies of each other. Digital camera chips and IMU, such as micro-machined gyroscopes and accelerometers, are now available off-the-shelf with good performance, and can provide robust estimates of self-motion as well as 3D scene structure, without any external infrastructure.