Investigating the Impact of Co-Learning Systems in Providing Customized, Real-Time Student Feedback

A current frontier in STEM education is the development of individualized learning environments that can help students learn challenging topics. The need is especially acute during active learning sessions and laboratory activities. This project will develop co-learning systems to enhance student performance in STEM laboratory activities. Co-learning systems consist of computer systems equipped with sensors. These systems can learn from humans and in turn interact with humans providing them with customized, real-time performance feedback. The lessons learned from the project can also be applied in other areas of STEM education such as the K-12 environment where co-learning robotic laboratory tutors could help alleviate some of the demands on the time and attention of science teachers.

The specific research objective of this Improving Undergraduate STEM Education (IUSE) project is to test the hypothesis that co-learning systems are able to enhance student performance in undergraduate STEM laboratory activities by providing them with customized, real-time performance feedback. The project will utilize existing commercial, off-the-shelf technologies in development of the systems. Sensor input will consist of audio, video, depth, skeletal, and 3D mesh data collected using a multimodal sensing device. The co-learning systems will be integrated into the laboratory environments to capture and translate data pertaining to STEM classroom environments and student interactions during laboratory activities. The type of data will include audio data pertaining to student verbal queries, skeletal data pertaining to student gesture patterns, and the content of student work. The project will investigate machine learning algorithms suitable for discovering knowledge pertaining to student learning during STEM laboratory activities. Work will assess the students' perception of co-learning systems and evaluate the ability of co-learning systems to improve performance during laboratories. The investigators will compare student learning outcomes between comparable student groups subject to different amounts of interaction with the co-learning systems. These results will provide information about the potential of co-learning systems to augment traditional teaching and provide an effective, automated, personal STEM learning environment.