Machine learning reveals the control mechanics of an insect wing hinge

Area of Science:

• Evolutionary Biology
• Biomechanics
• Insect Physiology

Background:

• Insect flight is a key factor in their evolutionary success, enabling them to diversify into millions of species.
• The insect wing is a novel structure, not evolved from limbs, attached by a complex biomechanical hinge.
• This hinge translates muscle oscillations into wing motion, regulated by specialized control muscles and sclerites.

Purpose of the Study:

• To elucidate the mechanical control logic of the insect wing hinge.
• To understand how steering muscle activity influences wing motion and aerodynamic forces.
• To integrate multi-disciplinary approaches for a comprehensive model of insect flight control.

Main Methods:

• In-situ imaging of steering muscle activity in flies using genetically encoded calcium indicators.
• High-speed 3D tracking of wing motion.
• Machine learning (convolutional neural network and encoder-decoder) to predict wing motion from muscle activity and sclerite function.
• Aerodynamic force quantification using a dynamically scaled robotic fly.
• Physics-based simulation of flight maneuvers.

Main Results:

• A machine learning model accurately predicts insect wing motion based on steering muscle activity.
• The study identified the role of individual sclerites in modulating wing motion.
• Robotic fly experiments quantified the aerodynamic effects of steering muscle activation.
• Simulations incorporating the hinge model replicate free-flying insect maneuvers.

Conclusions:

• The study reveals the intricate mechanical control logic governing the insect wing hinge.
• This research provides a sophisticated model for understanding insect flight, a crucial evolutionary adaptation.
• The findings highlight the wing hinge as a pivotal and complex skeletal structure in the natural world.