Intermittent Swimming Boosts Energy Efficiency of Fish-like Robots

Improving energy efficiency is crucial for extending robot operation time and reducing battery load, enabling lighter and more durable robotic systems. Nature has evolved energy-saving locomotion strategies through billions of years of natural selection, offering blueprints for robotic design.

Among aquatic locomotion modes, intermittent swimming, or bout-and-glide, is widespread in organisms from larval zebrafish to whales. This gait alternates between short bursts of active undulation and passive gliding with a streamlined posture, and is widely recognized as an energy-optimizing mechanism, making it a promising strategy for robotic control.

An international joint team from EPFL, Duke University, and Instituto Superior Técnico developed a larval zebrafish-inspired robotic platform called ZBot to systematically compare the performance of intermittent and continuous swimming.

The research focuses on four scientific questions: the neural control mechanism underlying intermittent swimming, the conditions for achieving higher energy efficiency, the impact of viscous flow regimes, and the mechanisms that lead to energy efficiency.

To validate the bioinspired energy-saving mechanism, the team designed ZBot, a 200-fold scaled-up version of a larval zebrafish, measuring 80 cm in length and weighing 2.8 kg. It replicates the zebrafish's morphology, segmented body, and center of mass distribution.

Its flexible tail includes six servomotor-driven segments to mimic natural undulation, while the head houses a central controller, high-precision cameras, and real-time power meters. Expandable sensor interfaces support diverse experiments, including visual-motor processing and vestibular research.

To explore the energy efficiency of intermittent swimming, the team developed a neurocomputational model of zebrafish neural circuits, centered on central pattern generators (CPGs), a bout gate, and ventral spinal projection neurons. The CPGs generate continuous oscillations, while the bout gate acts as a switch, triggering tail beats only when a threshold is reached, creating the natural active-bout and passive-glide rhythm.

By adjusting parameters, ZBot replicated multiple gaits, including slow straight swims, routine turns, and J-turns. The model was extended to reproduce visually guided optomotor responses in both the robot and its digital twin, simZFish.