Onboard/Offboard Sensor Fusion for High-Fidelity Flapping-Wing Robot Flight Data

In-flight data acquisition and flight testing for system identification of flapping-wing MAVs

“…In this section we outline an approach to combine the two previously presented data acquisition methods effectively, specifically for flapping-wing modeling and analysis, avoiding some of the aforementioned limitations. Further details can be found in [22].…”

“…An extended Kalman filter (EKF) was found to provide effective results [22], while being straightforward to implement. By suitable tuning of the measurement and process noise matrices it becomes possible to achieve accurate estimates for both the slow dynamics over long time scales, and the short-term fast dynamics.…”

“…As a reference, wind-tunnel forces measured with a sensitive load cell (ATI Nano17-Ti) for similar flight conditions are also shown (magenta). The test conditions were described in [21] and [22].…”

“…The typically available tracking and on-board data acquisition systems allow for estimation of the body attitude and velocity, as well as of the IMU sensor biases. A standard EKF can be used for this, with IMU accelerometer and gyroscope data as input variables and optical tracking data as output measurements [22], resulting in the following filter definition and equations: Measurement equations:…”

“…The present article addresses these specific requirements and presents recommendations and guidelines for the collection and respective post-processing of data obtained from free flight experimental techniques, based on prior testing made with different systems and vehicles across literature [9], [15]- [22]. In particular, it collects the best practices to perform free flight tests, compares different external tracking systems, compares them to on-board inertial measurement unit logging, and presents flight path reconstruction methods to accurately reconstruct and quantify the flight states and acting forces, while minimizing the error and uncertainty in the data.…”