Control of Synchronization Regimes in Networks of Mobile Interacting Agents
by Fernando Perez-Diaz (TUoS Department of Computer Science), Ruediger Zillmer (Unilever R&D, Port Sunlight) and Roderich Gross (TUoS Department of Automatic Control and Systems Engineering).
Dated: 5th May 2017
Congratulations to ACSE’s Dr Roderich Gross and his colleagues; Fernando Perez-Diaz and Ruediger Zillmer, on their latest paper, entitled 'Control of synchronization regimes in networks of mobile interacting agents', which has just been accepted by the Physical Review Applied Journal - a journal which looks as work at the intersection between Physics and Engineering.
The paper investigates synchronization in a population of mobile pulse-coupled agents with a view towards implementations in swarm robotics systems and mobile sensor networks. Previous theoretical approaches dealt with range and nearest neighbour interactions. In the latter case, a synchronization-hindering regime for intermediate agent mobility was found. In the present work, the robustness of the intermediate regime under practical scenarios was investigated.
In this work we experimentally confirm that the ability of multiple agents to synchronize their clocks may depend on the speed at which they move. If they move slow or fast, they synchronize well. However, counter-intuitively, if they move at a medium speed they are unable to reach consensus. The paper also sheds light on the underlying causes, and how existing synchronisation protocols can be adapted to allow agents to synchronize irrespective of their speeds.
Dr Gross, Department of automatic control and systems engineering
Dr Gross and his colleagues show that synchronization in the intermediate regime can be predicted by means of a suitable metric of the phase response curve. Further, they study more realistic K-nearest neighbours and cone of vision interactions, showing that it is possible to control the extent of the synchronization-hindering region by appropriately tuning the size of the neighbourhood. To assess the effect of noise, they analyse the propagation of perturbations over the network and draw an analogy between the response in the hindering regime and stable chaos. Their findings reveal the conditions for the control of clock or activity synchronization of agents with intermediate mobility. In addition, the emergence of the intermediate regime is validated experimentally using a swarm of physical robots interacting with cone of vision interactions.
The paper poses the question; How do we get in sync with each other?
This is through various sorts of feedback. Swarms of mobile pulse-coupled agents (such as robots) exhibit distinct synchronization regimes, which can be controlled by tuning the velocity of the agents, or their interaction rules, as Dr Gross and his colleagues show. Between slow and fast speeds lurks an intermediate zone where lock-step behaviour can be impeded—possibly on purpose, depending on an application’s goals.