Assisting in the Lab
Dan Gildea, associate professor of computer science, Henry Kautz, professor of computer science, and Jiebo Luo are using their expertise in computational linguistics, machine learning, and computer vision to build an automated wet lab assistant. This assistant can create a detailed written record of what each technician in a biology lab does and can detect when the activities are performed incorrectly.
While explicitly programming a system to accurately monitor all the different activities in a wet lab would be infeasible, the wet lab assistant “learns” about activities from the combination of written lab protocols and raw video of the laboratory. This technology will help scientists better record and increase the reproducibility of their experiments—a dire need in today’s biomedical research.