MATLAB-Simulink Auto-Grading for Self-Study Assignments and Exams
Summary of the project
The development of a MATLAB-Simulink based auto-grading system for self-study assignments and take-home exams in the Vehicle Control course (4AT080) has been initiated by Tom van der Sande from the Dynamics & Control department. The course structure combines 2-hour lectures with 2-hour guided self-study sessions where students tackle problems using MATLAB-Simulink. However, grading these take-home assignments has proven to be labor-intensive and error-prone, exacerbated by varied student approaches and issues with plagiarism. Reports, which are a major component of the course, often result in low-quality submissions, further complicating the grading process.
The proposal aims to address these challenges by setting up a MATLAB-based grader. The primary goal has been to provide each student with a unique problem with individualized parameters to prevent plagiarism. The grader will be designed to facilitate automatic feedback on students' solutions, thus streamlining grading and enhancing learning. By replacing report writing with direct submission of MATLAB models, the project is set to improve the quality and efficiency of the evaluation process. The first year involves the implementation of the grader for one take-home assignment, with subsequent years expanding its use to additional assignments and guided self-study tasks.
Aim of the project
The aim of the project has been to create an automated MATLAB-Simulink based grading system for the Vehicle Control course (4AT080) to address current grading inefficiencies and improve student learning. This system is intended to simplify the grading process by automating the assessment of take-home exams and guided self-study assignments. By offering students unique problems with specific parameters, the project aims to reduce plagiarism and enhance the accuracy of grading.
Additionally, the project seeks to implement automatic formative feedback, providing students with timely and individualized responses on their work. This innovation is expected to streamline grading, reduce manual errors, and allow students to focus more on solving problems rather than writing reports. The developed system is planned to be scalable to other similar courses, thus contributing to broader educational improvements within MATLAB-based assignments and assessments.
Results and learnings
In the first year of the project, the assignments given to the students have been evaluated on their applicability for use in the Matlab-Simulink grading scheme. Some minor inconsistencies have been corrected by the teaching assistants and all the questions that students had were recorded. Based on these questions, answers to the most common questions can be added to the Matlab-Simulink grader. In addition to that, the first attempts have been made in providing automated feedback. For now, only Matlab based assignments can be easily graded using the existing Matlab grader. Simulink still poses challenges since it causes either run time errors, or students have access to the final solution via the workspace they have access to.
For more information, please contact:
