Dresden 2020 – wissenschaftliches Programm
Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 16: Poster IV
BP 16.17: Poster
Dienstag, 17. März 2020, 14:00–16:00, P2/EG
Reinforcement Learning with Artificial Microswimmers — Frank Cichos1, Viktor Holubec2, and •Ravi Pradip1 — 1Molecular Nanophotonics, Peter Debye Institute for Soft Matter Physics, Leipzig, Germany — 2Charles University in Prague, Faculty of Mathematics and Physics
Artifical microswimmers are designed to mimic the motion of living microorganisms. The adaptive behavior of the latter is based on the experience they gain through the interactions with the environment. They are also subjected to Brownian motion at these length scales which randomizes their position and propulsion direction making it a key feature in the adaptation process. However, artificial systems are limited on their ability to adapt to such noise and environmental stimuli. A novel solution to this problem has already been demonstrated by incorporating machine learning algorithms: self thermophoretic artificial microswimmers are employed in a real-world environment controlled by a real-time microscopy system to introduce reinforcement learning. It has also been shown that the learning process in these noisy environments contributes to a decline in learning rate and varied optimal behavior. In addition, as a consequence of non zero delay between sensing and responding to external stimuli in such an environment an optimal velocity emerges for these microparticles which ensure the expected behavior. Therefore an effort to lower the current delay is made which will enable the particles to exploit the learned knowledge for a wider range of velocities.