Go Big or Go Home: Simulating Mobbing Behavior with Braitenbergian Robots

A research paper from the University of Waterloo explores how simple, biologically-inspired robots can exhibit complex group behavior. Led by Elaheh Sanoubari, the study 'Go Big or Go Home: Simulating Mobbing Behavior with Braitenbergian Robots' uses the Webots simulation platform to model 'mobbing'—a cooperative anti-predator strategy seen in animals. The team simulated Braitenbergian robots, which are minimalist agents with direct sensor-to-motor connections, facing a light source representing a predator. When encountering the threat, a robot emits a simulated mobbing call to summon allies; if enough robots respond, they collectively 'mob' the light, otherwise, they flee.

The research systematically tested two key variables: the effective range of the mobbing call (infinite, mid, and low range) and the size of the robot group (ten robots versus three). The results confirmed that both factors have a significant impact on the overall success rate of the mobbing behavior. A larger group and a longer call range consistently led to more effective cooperative attacks. The work, completed as a 2019 final project for the graduate course ECE 750 - Artificial Life: Embodied Intelligence, was recently posted to the arXiv preprint server.

This simulation-based investigation bridges concepts from artificial life and embodied intelligence. By demonstrating how simple reactive agents can achieve coordinated action through emergent communication, the study has clear implications for the field of multi-agent systems. The findings can inform the design of robust and decentralized control architectures for swarms of autonomous robots, where reliable group coordination without a central leader is essential for tasks like environmental monitoring or search and rescue.