How far can a 3pi go?

Hello Pololu forums,

I am thinking of purchasing a Pololu 3pi robot to coincide with my university project researching emergent behaviours.
In particular, inducing beneficial emergent behaviours and mitigating non-beneficial behaviours.

How complicated can I make the behaviour of the robot?
Can two or more 3pi’s perform cooperative tasks? Could I use 3 or more 3pi’s to form a heterogeneous system?
Could I implement reinforcement learning as a group?

Basically, Behaviour-based, Collective Behaviour, Adaptation, Reinforcement Learning, Evolutionary Learning, Genetic Algorithms, Self-organisation and Emergent Behaviour are the topics I would like to study. Could anybody please recommend suitable robots for this purpose? Or perhaps any simulation software that would accommodate all of these areas?

The 3pi looks great but I am unsure of how complex the behaviour can be. (Or if interactions are easy and suitable to model on the 3pi platform)

Thank you for your time,



The 3pi robot is designed specifically to excel in line-following and maze-solving tasks; however, that does not define the 3pi’s limitations, and it can make a great platform for more advanced robotic behaviors. You might find it helpful to see some project videos of what the 3pi can do.

I suggest you specifically consider the m3pi Robot with mbed NXP LPC1768 development board, which is a 3pi robot that has been expanded to serve as a slave base to an mbed development board. The mbed is a much more capable microcontroller than the ATmega328P on the 3pi, and it also has many available I/O lines that can be used to connect interesting peripherals like more sensors and wireless modules.

- Amanda