improved r&d cycles through data analysis and collaborative robots
Element Six (E6) wanted to accurately predict synthetic diamond material properties to improve their discovery and R&D cycles by utilising a solution that would enable automatic collection, integration, and storage of data from experiments carried out by a collaborative robot (CoBot).
Collection, storage, and analysis of CoBot collected data for synthetic super materials
"We have seen an increase in both our data quality and volume, enabling us to gain a deeper understanding of our solutions and testing processes... This has decreased the time taken to aggregate data on a daily basis and improved time efficiency in our R&D cycles.”
Sanchit Nardekar | Data Scientist | Element Six
T-DAB were engaged to develop a tool that enabled Element Six to automatically collect data from a collaborative robot, store and integrate it, and allow analysis of the data from both individual and aggregated experiments. Ultimately, the purpose of this was to facilitate the development of machine learning models.
T-DAB conducted a Discovery and Design exercise to assess the data available to E6 and identify a suitable architecture platform using AWS and Azure to support their operations.
Through the discovery exercise, legacy data in Excel, images, and flat files were loaded into a SQL server instance using an ETL process. This allowed T-DAB to perform an exploratory data analysis (EDA) and machine learning Proof of Concept (PoC) to understand and recommend what data should be collected by the CoBot collection framework.
During the development of the CoBot collection framework, this ETL process ran as a windows service. It continued to harvest manually collected data. The ETL process was integrated with a legacy Lotus Notes database which contained the experimentation data that had previously been printed out as a PDF to be manually followed by the material scientist. This manual process was then presented to the user as a webform to remove the need for re-entering data. The data collected from IoT sensors on the CoBot was then integrated into the data entry process.
the tools involved
7x Increase In Data Collection
Up To 10-15x Using The Cobot
3x Increase In Data Quality
hours to seconds
The collaboration helped E6 to reduce aggregation time for analysis between different tests from hours to seconds via a Power BI dashboard built using the clean data.
Element Six acted as domain experts throughout the project with T-DAB providing guidance and steering the development of the project. This collaboration resulted in a 7x increase in data collection, with the potential of up to 10-15x using the CoBot, as well as a 3x increase in data quality.
In terms of analysis, it helped E6 to reduce aggregation time for analysis between different tests from hours to seconds via a Power BI dashboard built using the clean data (the reduction in aggregation time of the data, from around 4 hours to 0 hours was the most prominent metric).
This in turn helped deliver vital testing information to the R&D team faster and with greater traceability and accuracy.
T-DAB always emphasise the importance of collaboration, particularly between domain and technical experts, and in the case of such a broad brief, teamwork was of particular importance.
A core requirement for E6 was to build up an in-house data science team. Therefore, while T-DAB lead the technical delivery of the project, we also provided lasting legacy services including:
- Integration of E6 junior data scientists and engineers to the project team, working alongside T-DAB. Full integration into daily stand ups, weekly reviews, and team collaborative working tools provided by T-DAB.
- Provision of weekly technical help clinics for E6 junior engineer; introducing Data Engineering principles including SQL server and ETL, which have since been used across digital transformation projects.
- Provision of weekly formal data science tutorial classes for the client junior data scientist; introducing the theory behind many Machine Learning methods, particularly Bayesian Statistics and Convolutional Neural Networks, which have been used to improve the accuracy of a CNN classifier by 5%.
T-DAB provided the technical expertise to drive and deliver this element of the project, while ensuring the proper exchange of knowledge to allow E6 to gain understanding and make best use of the tool. This ensured immediate and lasting value. Close collaboration with the client combined with an agile development framework allowed us work with in-house support teams to deliver the project into production. There was a handover period where T-DAB continued to support the application, and E6 has now been completely self-sufficient for the last 6 months.
Overall, the project greatly improved data input, quality and storage, and serves as a foundation for further digital additions, as well as setting a precedent for the benefits data can bring when harnessed properly.
“As an R&D centre, our data is often very varied. Thus storing our data accurately is important for traceability and analysis purposes. T-DAB were able to aid us with structuring our data, integrating it with our existing frameworks to provide bespoke software and database solutions.
As a result, we have seen an increase in both our data quality and volume, enabling us to gain a deeper understanding of our solutions and testing processes, as well as allowing a dashboard to be built. This has decreased the time taken to aggregate data on a daily basis and improved time efficiency in our R&D cycles.”
Sanchit Nardekar | Data Scientist | Element Six
Element Six (E6) is a global leader in the design, development and production of synthetic diamond and tungsten carbide supermaterials.
Since 1959, their focus has been on delivering extreme performance through the development of innovative, cutting-edge synthetic diamond and tungsten carbide solutions. As well as being the planet’s hardest material, diamond’s extreme and diverse properties give it high tensile strength, chemical inertness, broad optical transmission and very high thermal conductivity.