Quantifying the biomimicry gap in biohybrid robot-fish pairs

Vaios Papaspyros,Guy Theraulaz,Clément Sire,Francesco Mondada

2024-03-18

Abstract:Biohybrid systems in which robotic lures interact with animals have become compelling tools for probing and identifying the mechanisms underlying collective animal behavior. One key challenge lies in the transfer of social interaction models from simulations to reality, using robotics to validate the modeling hypotheses. This challenge arises in bridging what we term the "biomimicry gap", which is caused by imperfect robotic replicas, communication cues and physics constraints not incorporated in the simulations, that may elicit unrealistic behavioral responses in animals. In this work, we used a biomimetic lure of a rummy-nose tetra fish (Hemigrammus rhodostomus) and a neural network (NN) model for generating biomimetic social interactions. Through experiments with a biohybrid pair comprising a fish and the robotic lure, a pair of real fish, and simulations of pairs of fish, we demonstrate that our biohybrid system generates social interactions mirroring those of genuine fish pairs. Our analyses highlight that: 1) the lure and NN maintain minimal deviation in real-world interactions compared to simulations and fish-only experiments, 2) our NN controls the robot efficiently in real-time, and 3) a comprehensive validation is crucial to bridge the biomimicry gap, ensuring realistic biohybrid systems.

Robotics,Machine Learning,Quantitative Methods

What problem does this paper attempt to address?

The paper attempts to address the issue of quantifying the "biomimicry gap" in bio-hybrid systems. Specifically, the researchers used a biomimetic lure and a neural network model to generate biomimetic social interactions and validated the effectiveness of these biomimetic systems through a series of experiments. The main objectives of the paper include:

1. **Quantifying the Biomimicry Gap**: The bio-hybrid system in the study consists of a real fish (Hemigrammus rhodostomus) and a robotic lure. By comparing the interactions between the real fish and the robotic lure, as well as interactions in a simulated environment, the researchers aim to quantify the gap between the system's performance in practical applications and the ideal scenario.

2. **Validating the Model's Effectiveness**: The study utilized a Deep Learning Interaction (DLI) model to control the robotic lure and evaluated the model's performance in the real world through long-term experiments (approximately 11 hours). The results showed that the DLI model could effectively control the robotic lure in real-time environments.

3. **Comprehensive System Validation**: To ensure the system's authenticity, the researchers conducted multi-level validations, including individual behavioral differences, collective dynamics, and long-term behavioral patterns. These validations help bridge the gap between simulation and reality, ensuring the reliability of the bio-hybrid system.

In summary, the paper aims to explore how to achieve more realistic biomimetic interactions in bio-hybrid systems through detailed experimental design and data analysis, thereby better understanding and simulating the social behavior of fish.