With the rise of data science and machine learning in science the desire for data has never been higher. One solution to this proposed by many is through automation and the development of high-throughput, remotely controllable instruments. This is resulting in an education challenge; how does one learn to automate a remote instrument, and perform the subsequent analysis of the data stream? We developed a lightweight Raspberry Pi powered instrument to provide an on-ramp to the automation super-highway. Our instrument consists of an RGB LED with three inputs that are the red, green and blue LED levels. There is a light sensor that measures the intensity of light at 8 different wavelengths. We call the instrument Claude-Light, and it can be built for under $200. The instrument is accessible through a web browser with fillable html forms, and via an API for programmatic access through shell commands, Python scripts, or Jupyter notebooks. At the time this abstract was prepared Claude-Light has run over 150K experiments around the world.
We developed several virtual instruments for different learning purposes. For example, the "GreenMachine" (https://claude-light.cheme.cmu.edu/gm) has only one controllable parameter for the green RGB channel, and the instrument returns only one output, the intensity of measured green light. The "RGB" machine (https://claude-light.cheme.cmu.edu/rgb) has three inputs for the red, green and blue LED channels, and three outputs for the intensity of light at red, green and blue wavelengths. Finally, an API machine allows access to all the inputs, and all the outputs.
This simple instrument can be used for a broad range of educational activities that range from statistics and reproducibility studies, activities with design of experiments, explorations of data science and machine learning, and studies of data structures and automation algorithms. We have used the "GreenMachine" instrument several times in the class room with first year students in collaboration with faculty teaching Introduction to Chemical Engineering. We have also used the instrument in research, for example exploring automation technology with large language models, and in solving inverse problems.
In this talk we will present Claude-Light. We will show applications that could be used across a 4 year STEM program to learn about statistics, design of experiments, data science and automation technology. Additionally, there are opportunities to learn about instrument design, web programming, and API design. Claude-Light clearly does not solve all the challenges in learning about automation and high-throughput data production and analysis, but it does provide a stepping stone towards more advanced instruments.