Welcome to our tutorial on smart chemical robots
18th September 2017, 14h-17h30, https://goo.gl/maps/jyeuYzuDRXC2
We believe principles from developmental robotics can be of benefit when designing curious robotic assistants for lab environments. In this tutorial, we will introduce and showcase examples from robotic assistants in our chemistry lab and how they help us characterize our chemical systems more effectively. In the video below, we showcase our oil-in-water droplet system that can display a fascinating range of behavior.
Our robots are able to perform physical or chemical experiments with the aim to explore the states such complex systems can exhibit, a challenge alike sensory-motor exploration in humans and robots. The tutorial will be of special interest for researchers interested in sensory-motor exploration and its application outside of robotics and developmental sciences.
Our aim is to highlight the potential benefits of interdisciplinary work at the interface of physics, chemistry, robotics and AI.
For more details and background information, please read the tutorial proposal.
The tutorial will take place the 18th September 2017 from 14h to 17h30 during the http://www.icdl-epirob.org/ conference. The conference will be held at the Congress Center of the Instituto Superior Tecnico (maps: https://goo.gl/maps/jyeuYzuDRXC2), in Lisbon, Portugal.
Provisional schedule is as follow:
We are looking for a senior researcher willing to give a 20 minutes perspective talk on the subject and its articulation with the DevRob community. If you are interested or know someone that might be, please contact me at email@example.com.
- Juan Manuel Parrilla Gutierrez, et al. “Evolution of oil droplets in a chemorobotic platform.” Nature communications (2014)
This is the first work published by the group in that area. It demonstrates how we can artificially evolve oil-in-water droplets to maximize different behaviors type, such as speed or division.
Link to journal: https://www.nature.com/articles/ncomms6571
- Vasilios Duros, Jonathan Grizou, et al. “Human vs Robots in the Discovery and Crystallization of Gigantic Polyoxometalates”. Angewandte Chemie (2017)
In this work, we use active learning classification methods to define the next experiments to perform to characterize faster a crystalization ‘zone’ of a new polyoxometalate cluster. Importantly, we compare our approach with human experimenters showing variability between each experimenters. Overall active learning was beneficial for exploring more efficiently our crystal system.
Code and data: https://github.com/croningp/crystal_active_learning
Link to journal: http://onlinelibrary.wiley.com/doi/10.1002/anie.201705721/abstract
- Juan Manuel Parrilla Gutierrez, et al. “Artificial Evolution of Droplet Protocells in a Chemical Robot with Configurable Environments Leads to Phenotypic Plasticity”, Nature communications (2017)
This paper explores the impact of the physical environment on the droplets behaviors. To this end, we encapsulated the functionalities of our chemical robots into a 3D printed device, allowing us to modify the shape of the arena.
Link to journal: https://www.nature.com/articles/s41467-017-01161-8
- Laurie J. Points, et al. “Artificial Intelligence Exploration of Unstable Protocells Leads to Predictable Properties and Discovery of Collective Behaviour”, PNAS (2018)
This paper presents our efforts to understand the underlying physical and chemical processes at work in our droplet system. We also explore the impact of the chemical environment of the droplets on their behaviors.
Link to journal: http://www.pnas.org/content/early/2018/01/09/1711089115
- Jonathan Grizou, et al. “Curious robotic assistant unveils the surprising temperature sensibility of oil-in-water droplet motion”, In preparation
Inspired by recent trends in developmental sciences, we describe a curious robotic assistant able to explore an unknown system and unveil the range of states it can exhibit for the scientist to later analyze. The goal babbling enables to observe that temperature lead to abrupt and non-linear changes in the droplet motions, a phenomenon never reported before in the literature
Job opportunities are available for prospective PhD students and PostDocs with a strong desire to work in this innovative and challenging research program. The group employs chemists, roboticists, statisticians and computer scientists specialized in machine learning. We encourage interested researchers to attend to the tutorial and contact us for more information.
The work presented is undergone by the Chemobot team within the Cronin Group, in the School of Chemistry at the University of Glasgow.
The tutorial is organized by:
- Jonathan Grizou, team leader of the Chemobot team (4 PostDocs, 3 PhD students, 1 Intern), background in Develepmental Robotics.
- Laurie Points, PhD student in the Chemobot team, background in Chemistry.
- Lee Cronin, Professor and head of the Cronin Group (55 members), his group is pioneering research at the intersection of Chemistry, Robotics and Artificial Intelligence.