Outdoor Autonomous Navigation of a Tailless Flapping-Wing Micro Aerial Vehicle Using Low-Cost MEMS Sensors and GNSS...

Bio-inspired tailless flapping-wing micro aerial vehicles (FWMAVs) offer advantages of hovering flight, small size, and high maneuverability, showing potential for confined space missions. Outdoor autonomous flight remains challenging for tailless FWMAVs because of their strict payload and onboard-computing constraints. These constraints limit the use of high-precision sensors and computationally intensive perception methods, making reliable onboard state estimation difficult in outdoor environments. To address this issue, this paper presents an onboard autonomous flight scheme for a 31 g insect-inspired tailless FWMAV with a wingspan of 33 cm. Low-cost MEMS sensors and a GNSS module are used to obtain onboard motion and position information. An error-state Kalman filter (ESKF) fuses this measurement information to provide continuous estimates of the vehicle state. Based on the estimated state, a cascaded PID controller regulates the vehicle position and motion in a closed loop. All sensing, state estimation, and control functions are performed onboard without relying on external localization infrastructure. Outdoor experiments showed that the fused position estimates remained consistent with independently recorded real-time kinematic (RTK) GNSS measurements. The vehicle achieved autonomous position holding and stable flight along predefined paths under open-sky GNSS conditions. During three independent 20 m straight-line trajectory tracking flight trials, the average maximum lateral and vertical errors were 0.28 m and 0.10 m, respectively. A V-shaped altitude-varying trajectory tracking flight further demonstrated simultaneous horizontal and vertical motion, with a three-dimensional estimated-state-to-reference root mean square error (RMSE) of 0.0601 m. These results demonstrate the feasibility of fully onboard outdoor autonomous flight for tailless FWMAVs with sub-0.5 m wingspan under open-sky GNSS conditions.