Synchronization of Swarm Robots Using Pulse Coupled Oscillators
Abstract
Our research was focused on a new and upcoming model called Pulse Coupled Oscillators or PCOs. As the swarm robotics field grows, PCOs are being considered as an efficient and decentralized way to synchronize robot headings while also sending minimal data over networks. Oscillators are models that spin, and when they complete a rotation they send a pulse to the other oscillators, where an algorithm will be run to determine what the oscillator needs to change its phase value to. During our research we explored two methods: the Mirollo Strogatz algorithm and the Phase Response function (PRF) method, both of which were coded based on research papers. We then used Raspberry Pis’ as our PCOs and ran the code on three of them at a time. Our research showed that the PRF method was quicker than the Mirollo-Strogatz algorithm, but was way less reliable as there was a chance that they would never synchronize at all, on the other hand the Mirollo-Strogatz method was slower but would always synchronize. These results are what our mentor Timothy Anglea found as well, from there he used machine learning to combine and maximize the two methods and made one that was both fast and reliable. In the future, PCO algorithms and machine learning will continue to be combined and studied in order to further the control of swarm robots.
Synchronization of Swarm Robots Using Pulse Coupled Oscillators
HSS 209
Our research was focused on a new and upcoming model called Pulse Coupled Oscillators or PCOs. As the swarm robotics field grows, PCOs are being considered as an efficient and decentralized way to synchronize robot headings while also sending minimal data over networks. Oscillators are models that spin, and when they complete a rotation they send a pulse to the other oscillators, where an algorithm will be run to determine what the oscillator needs to change its phase value to. During our research we explored two methods: the Mirollo Strogatz algorithm and the Phase Response function (PRF) method, both of which were coded based on research papers. We then used Raspberry Pis’ as our PCOs and ran the code on three of them at a time. Our research showed that the PRF method was quicker than the Mirollo-Strogatz algorithm, but was way less reliable as there was a chance that they would never synchronize at all, on the other hand the Mirollo-Strogatz method was slower but would always synchronize. These results are what our mentor Timothy Anglea found as well, from there he used machine learning to combine and maximize the two methods and made one that was both fast and reliable. In the future, PCO algorithms and machine learning will continue to be combined and studied in order to further the control of swarm robots.
