Understanding Collective Motion

A recent discovery at the University Of Warwick has applications in a host of fields such as robotics, drone swarms, farming and even CGI graphics, where creating realistic swarms is seen as a gold standard.

Scientists at the University of Warwick used flocks of birds as a model to shown that birds of a feather will indeed flock together to maximise the information they have access to and to give them the most future options. They discovered that similarly, people’s choices like where to work or live are made influenced by the need to keep options as open as possible – and the more we co-operate together, the more opportunities are available to us.

The discovery by Henry Charlesworth and his supervisor Professor Matthew Turner published on 15 July in Proceedings of the National Academy of Sciences and provides a clue to the emergence of social co-operation in animals by explaining how individuals gain greater advantages by working in groups. The research was partially funded by the Engineering and Physical Sciences Research Council (EPSRC), part of UK Research and Innovation.

The researchers sought to gain a better understanding of collective motion, like that seen in a flock of birds, a herd of animals, an insect swarm or a human crowd.

They created a computer simulation, using bird flocks as a model, in which the ‘birds’ perceived a visual representation of the world around them, as if through a simple retina. They then programmed the models with an algorithm based on the principle of Future State Maximisation (FSM), so the ‘birds’ would move to maximise the number of different visual environments that they expect to be able to access in the future.

The way they moved together resembled animals in several ways, including cohesion (they stick together), co-alignment (they fly in roughly the same direction as their neighbours) and collision suppression, none of which were specifically programmed into the model. This demonstrates that there is a fundamental advantage to the ‘birds’ in working together. Dynamic trajectories emerge spontaneously for agents moving according to Future State Maximization i.e. maximizing their control over future visual states that they can access.

Professor Matthew Turner, from the University Of Warwick Department Of Physics, said: “We adopted a hypothesis that birds are agents that want to maximise their future freedom, and then we asked what the consequences are of that. It looks like it generates dynamics that are extremely similar, even at the quantitative level, to a bird flock.”

The algorithm is similar to ‘tree searches’ that have been used for a number of years in applications like chess programs. Chess algorithms would build tree searches of future lines of play and then select those lines that give them the maximum future options, among other factors.

This latest research also suggests that this principle may be a fundamental tool for information processing agents and perhaps help to define intelligence itself. The findings were compiled in paper titled ‘Intrinsically motivated collective motion’ and published in PNAS.

Citation: Intrinsically motivated collective motion, Henry J. Charlesworth, Matthew S. Turner, Proceedings of the National Academy of Sciences Jul 2019, 201822069; DOI: 10.1073/pnas.1822069116, https://www.pnas.org/content/early/20