Flapping-wing MAVs move closer to indoor ops as gust errors drop 53%

Key facts

• Chiba University study reports a bandwidth-constrained disturbance observer for flapping-wing MAVs to improve gust rejection.
• Testing used a 103 g commercially available Flapping Nimble+ robot; authors identify non-minimum-phase behaviour as a key control challenge.
• Reported results: 53.1% reduction in X-axis position error and ~28% reduction in total 3D position error under wind disturbance.

3 minute read

Flapping-wing micro aerial vehicles have long promised access to operating environments that are structurally hostile to conventional multirotors—collapsed buildings, narrow pipes, dense vegetation and other confined spaces where rotor strikes, entanglement and downwash are acute constraints. Their persistent blocker has been wind and disturbance sensitivity: low mass and oscillatory lift generation make them prone to being displaced faster than typical controllers can correct without inducing instability.

According to the cited Chiba University report (with the underlying study published in Control Engineering Practice), the team attributes much of the control difficulty to non-minimum-phase behaviour observed on a 103 g Flapping Nimble+ robot. In practice, when the controller commands a rapid horizontal correction—exactly the type of manoeuvre required to counter a gust—the vehicle initially moves in the opposite direction before recovering. This inverted transient can cause “fight the air” oscillations when controllers are tuned aggressively, while conservative tuning allows unacceptable drift and collision risk in tight quarters.

The proposed mitigation is a custom bandwidth-constrained disturbance observer, intended to estimate wind forces in real time while respecting the platform’s peculiar transient response. By constraining observer bandwidth—effectively moderating how quickly the estimator reacts—the approach aims to sit between the two failure modes: it avoids overreaction that amplifies oscillations, yet remains responsive enough to arrest drift.

Reported performance improvements are significant for this niche: a 53.1% reduction in X-axis position error and an approximately 28% reduction in overall 3D position error under wind disturbance. While this does not resolve enduring limitations of FW‑MAVs (payload, endurance, and manufacturability at scale), it strengthens the case that control, rather than only mechanics, is a tractable path to operational utility.

For European defence and civil protection users, the implication is less about replacing multirotors and more about opening capability gaps: indoor ISR in urban operations, post-blast or collapse reconnaissance, and inspection of critical infrastructure where collision tolerance and reduced rotor hazard are prized. European OEMs and integrators focused on protected-rotor designs (e.g., caged multirotors) may view FW‑MAVs as complementary rather than competing, but the control advance suggests the technology is moving from laboratory novelty toward more repeatable performance in disturbed airflow—an important precondition for procurement-grade indoor autonomy and assured operations in confined European built environments.

Source: The Drone Girl