EPFL researchers build 80cm robotic zebrafish to decode brain control

Researchers at EPFL have developed Zbot, an 80-centimeter robotic zebrafish that combines real-time neural circuitry with physical embodiment to study how brain-body-environment interactions drive animal behavior. The breakthrough demonstrates how artificial organisms can decode complex biological systems and advance both neuroscience understanding and robotics design.

What you should know: The research bridges the gap between isolated neural studies and real-world animal behavior by creating a complete artificial organism.

• Scientists used neural network data from live larval zebrafish, captured through calcium imaging techniques that record individual brain cell activity.
• The team first built a virtual zebrafish simulation that successfully replicated natural swimming behaviors like the optomotor response—a reflex that helps fish maintain position against water currents.
• Zbot translates these digital discoveries into physical reality, equipped with dual cameras as eyes and sophisticated motors that replicate segmented tail movements.

The big picture: Traditional neuroscience struggles to understand how brains function within living bodies interacting with dynamic environments, typically studying neural circuits in isolation under controlled laboratory conditions.

• This embodied approach reveals that the majority of neural signals driving behavioral responses originate from a focused region of the retina, a previously underappreciated insight.
• The computational model predicted two novel neuron types necessary to explain responses to complex visual stimuli, paving the way for future physiological experiments.

How it works: Zbot operates using the same neural control circuits embedded in computer simulations, allowing real-world validation of digital discoveries.

• The robot successfully demonstrated the optomotor reflex during tests in Lausanne’s Chamberonne River, swimming upstream and maintaining station despite turbulent flow patterns.
• This performance showcases how intrinsic circuit dynamics converge robustly to reorient organisms against environmental perturbations—a fundamental survival mechanism.

Why this matters: The research validates that visual inputs alone are sufficient for locomotor compensation in zebrafish, isolating sensation and motor control mechanisms impossible to study in living animals.

• Unlike traditional animal experiments constrained by biological limitations, the controlled simulation and robotic platform allows researchers to systematically isolate and manipulate variables.
• The approach enables scientists to determine the minimal circuits required for specific behaviors by selectively “turning off” pathways—something impossible in live organisms.

Competitive landscape: EPFL’s BioRobotics Lab is sharing their simulation platform and robot designs as open-source resources, inviting global scientific collaboration.

• The integration of biomechanics with neural control in natural habitats marks a transformative step beyond classical robotics that operate in simplified contexts.
• This work positions embodied artificial models as valuable testbeds for evolutionary and comparative studies on sensorimotor adaptation.

What’s next: Further developments at the BioRobotics Lab focus on unraveling complexities of zebrafish swimming patterns and multisensory coordination.

• The research sets the foundation for a new era of embodied neuroscience that bridges molecular insights with robotic reenactment of animal behavior.
• Applications extend to developing autonomous robots capable of sophisticated, adaptive behaviors in fluid environments.