Summary
Counting the wingbeats of small flying drones by watching their motion is tricky because the wings move very fast. The authors tested three different computer models that look at the movement captured by virtual cameras to count wingbeats accurately. They found that all three models, including one inspired by brain cells and one that uses attention mechanisms, were similarly good at counting wingbeats at different distances. The study shows that it is possible to count wingbeats precisely using this type of visual information, although real-world tests and efficiency in hardware are not yet addressed.
What this means in practice
- •For drone developers: Count wingbeats accurately in simulation to improve control and monitoring of flapping-wing drones using optical flow input.
- •For robotics vision engineers: Implement wingbeat detection from camera motion fields to analyze fast periodic motions in flying robots or bio-inspired machines.
Tested on simulated data.
Abstract
Visual monitoring of flapping-wing vehicles requires distinguishing individual wingbeats from motion strength and average frequency. This paper presents a controlled MuJoCo evaluation of wingbeat counting from signed optical flow observed by virtual cameras mounted on Crazyflie vehicles. Three flapping-wing models were recorded at optical distances of 1.5 and 3.0 m, producing 1,440 clips from 240 paired scene configurations with a scene-level 3:1 training-test split. A common spatial convolutional encoder was combined with a causal temporal convolutional network, a recurrent leaky integrate-and-fire spiking network, or causal self-attention. Each model predicted phase and activity, followed by the same directed-crossing event counter. The six existing convolutional models were retained, and all twelve new models were frozen before their test predictions were generated. Exact-count accuracies at 1.5 m were 96.67%, 95.00%, and 96.67%, respectively; at 3.0 m they were 94.44%, 92.22%, and 95.00%. All paired scene-bootstrap intervals for differences in exact-count accuracy included zero. Seven far-distance spiking-model clips had correct totals despite event-timing mismatches, demonstrating why total-count and event-level measurements must be reported together. The results support the feasibility of causal optical-flow counting in the tested setting and identify boundary-sensitive errors. They do not establish an architecture ranking across repeated training, real-flight robustness, or hardware efficiency.