Return to base and precision docking of an indoor drone – SpotReturn

The SpotReturn project focuses on the problem of GPS-denied navigation of indoor drones. It targets two scientific challenges: (1) the estimation of the distance traveled for a return to base by using a visual perception consisting of visual cues of optical flow, and (2) the precise docking of the drone by using a sensor with the following characteristics: accuracy, low latency, low power consumption, insensitivities to interferences, robust to external constraints (dust, smoke, varying lighting conditions).

We will estimate the distance traveled by the exploration drone so that it finds its way back despite disturbances (terrain, air current). We will validate on a drone the visual odometer called SOFIa that estimates the distance traveled accurately based only on a minimalistic vision inspired by the vision of insects. We will show that the optical flow perceived during the flight to which a visual scaling is applied allows to have an estimation of the distance traveled and thus to return to the docking station.

We will realize the precise docking on a surface of 2cmx2cm of an indoor exploration drone (20cm of scale). Based on the InLock sensor, we will estimate with precision the relative position of the drone with respect to the docking station in an environment without satellite signal and direct communication. The drone will also fly in difficult conditions (inhomogeneous lighting, presence of smoke, dust, electromagnetic waves).

We will experimentally validate the return to the base station and the precision docking with a fully actuated hexarotor, which will intrinsically stabilize the orientation of the sensors.