Achieving the United Nations’ Sustainable Development Goals (SDGs) hinges on accelerating scientific discovery. It is the only way to deliver new technologies, products, and services like clean energy technologies, affordable healthcare, sustainable agriculture practices, and more. However, its pace is limited by manual research undertaken in disconnected laboratories around the world with siloed assets and data. Therefore, to dramatically accelerate scientific discovery, we must digitise, automate, and connect these research laboratories.
How can we connect research laboratories around the world?
Connecting labs around the world could dramatically accelerate chemical discoveries, reduce lab operating costs, increase sustainability, and improve collaboration. However, there are major challenges to overcome. To achieve this goal, we must:
- Orchestrate and harmonise diverse tools, equipment, and methodologies across different labs.
- Improve the generation, integration, and sharing of data among research entities.
- Ensure data integrity and adopt FAIR (Findable, Accessible, Interoperable, and Reusable) principles to enhance data quality and reproducibility.
- Develop an infrastructure that supports transparency and collaboration in scientific research.
The World Avatar emerges as a groundbreaking solution to these challenges. It is an open digital ecosystem which uses a dynamic knowledge graph to connect the behaviours of various entities across different scientific domains. By structuring and linking real-world entities and concepts, the World Avatar enables applications that transcend traditional boundaries, facilitating cross-scale and cross-domain research collaborations.
Using the technology behind the World Avatar, we are creating a live, connected digital twin of our lab. We call this the “Digital Lab Framework”, where all data, software, hardware, and workflow are connected and interoperable through ontologies in knowledge graphs. Then, autonomous computational agents within the knowledge graph can automatically pull insights from the connected data and streamline the workflow. These knowledge graphs are then distributed over the internet using semantic web technology. We have also implemented authorisation and authentication mechanisms to ensure the assets are only accessible to the correct users and agents.
How are we implementing our Digital Lab Framework?
Our Digital Lab Framework is implemented in several stages:
- We define ontologies to capture the entities and conceivable concepts involved in a chemical laboratory.
- We design autonomous computational agents to act as executable knowledge components that manage the data and material flow within one laboratory and across laboratories.
- We create digital twins of laboratories and connect them via a dynamic knowledge graph to carry out these goals. The knowledge graph acts as a living copy of the laboratories and enables real-time investigation. All provenance from the agents' operations is recorded in the knowledge graph to ensure FAIR This also allows for backtracking historical events.
- The autonomous agents derive and execute goals, breaking down the high-level abstract goals of scientific research into smaller actionable tasks.
This methodology offers a holistic view of lab automation which considers all aspects of the laboratory beyond a single device or experiment. This enables the digitisation and automation of a wide range of different tasks that can be captured in the following roles:
- A “Digital Research Scientist” plans, conducts, and analyses experiments based on chemical knowledge, access to live data, and control of equipment (such as measurement devices and robots).
- A “Digital Laboratory Manager” monitors all ongoing activities in the lab to ensure safe operations, assigns workspaces, and tracks equipment.
- A “Digital Facility Manager” ensures stable environmental conditions and improves resource efficiency by monitoring utilities and controlling airflow, temperature, etc.
This is a “systems engineering” approach to lab automation that fundamentally differs from contemporary platform solutions. Most platform solutions try to be compatible with additional devices and processes on an ad-hoc basis (bottom-up approach). Instead, we engineer knowledge architectures that can represent abstract concepts such as goals which then propagate all the way down to a chemistry level (top-down approach).
We also consider the human-in-the-loop with this approach for two reasons: First, humans will be part of research laboratories for many years. Hence, we cannot jump to completely automated but must bridge the ‘interim technology gap’. Second, for us, automation is not an end in itself. By using a goal-driven approach, we may find that humans spend more time thinking of the correct goals to impose in the World Avatar.
What have we achieved?
With our Digital Laboratory Framework, we have already demonstrated the practical application of our framework by linking two robots in Cambridge and Singapore to achieve a collaborative closed-loop optimisation for a pharmaceutically relevant aldol condensation reaction in real-time. Our knowledge graph evolves autonomously while progressing towards the research goals set by the scientist. The two robots effectively produced a Pareto front for the cost-yield optimisation problem over the course of three days of operation.
By following a goal-driven approach, our Digital Laboratory Framework can formulate more high-level and complex goals that include resource efficiency. For example, it can ensure an automated experiment is stopped if the expected knowledge gain is getting too small in comparison to the expected resource consumption.
In addition, by taking a holistic view of lab automation/digitisation, our model considers the surrounding infrastructure and resource consumption as well. For example, it factors in the energy needed for the fume hood under which an experiment is carried out. This means that we can improve the efficiency of the laboratory and chemical research process in general.
Finally, the technology demonstrated in this work can be used to connect other laboratories in the world to facilitate collaboration between research teams. This will eventually bring us to the critical mass of a globally integrated autonomous experimentation system, where the size and degree of interconnectedness greatly multiply the impact of each research robot’s contribution to the network.
Integrating and automating research laboratories is crucial for advancing science and tackling global issues. The World Avatar achieves this using a dynamic knowledge graph to create a Digital Lab Framework. With the Digital Lab Framework, geographical boundaries are eliminated, enabling a seamless exchange of data and materials between laboratories. This breakthrough fosters collaboration and dramatically accelerates scientific progress. We have showcased the potential of this approach which paves the way for more efficient and impactful scientific endeavours.