Multi-robot replication of ant collective towing behaviours

Area of Science:

• Robotics
• Collective Behavior
• Animal Behavior

Background:

• Ants exhibit complex cooperative transport behaviors.
• A previous observation showed that the steady-state transport speed of Novomessor cockerelli ants decreases with increasing team size.
• The mechanistic basis for this phenomenon remained unclear.

Purpose of the Study:

• To investigate the mechanisms underlying cooperative transport in ants.
• To explain the observed decrease in transport speed with increasing team size.
• To validate hypotheses using a robotic swarm model.

Main Methods:

• Two computational models of collective towing were developed: one with variable forces and differing maximum speeds, and another with identical speeds and intermittent forces.
• Statistical analysis of Novomessor cockerelli ant data was performed to determine which model best fits ant behavior.
• Decentralized controllers based on the models were implemented on teams of mobile robots (2-4 robots).
• A real-time reinforcement learning algorithm was used for the controller simulating variable forces.

Main Results:

• Ant data statistically supported the model where steady-state transport speed is determined by the maximum speed of the slowest teammate.
• The data were inconsistent with the model predicting constant speed regardless of team size.
• Robotic swarm experiments successfully reproduced the ant-observed relationship between team size and transport speed using the first model's controller.
• The second model's controller on robots showed the predicted invariance of transport speed with team size.

Conclusions:

• The study validates a mechanistic hypothesis for cooperative transport in ants, identifying the slowest member's speed as the limiting factor.
• Robotic swarms serve as effective platforms for testing and validating hypotheses about biological collective behaviors.
• This research highlights the synergy between biological observation and robotic implementation in understanding complex systems.