Abstract:

Nowadays, many educational organizations find it challenging to correct practical experiments in remote laboratories, where in these experiments, the student would like to have speedy feedback since knowing where he went wrong or where he went right is very important to continue the teaching and learning process. To reduce the problem of performing the automatic correction of remote experiments in laboratories, an application was developed that stores data from a correct execution of the activity requested from the student, and when the student is going to execute this activity, an automatic correction algorithm will be applied to give the feedback of assessment for the student. This algorithm uses digital twin technology to acquire the input and output data during the experiment, and according to the images collected from the correct execution, a verification of the movements of the didactic task is mounted. This correction verifies if the digital twin data is accurate and what the position of the elements of the didactic plan is. The focus of this work is to present a system capable of automatically correcting the didactic plan based on a correct execution at the time of the activity registration, allowing the student to carry out his experiment and the correction with the evaluation automatically. The system consists of three stages: learning how to execute the experiment correctly, storing the images that characterize the state changes in the didactic experiment, and feedback with the student's evaluation. The learning stage aims to acquire data from the experiment in operation based on the initial registration of tags that identify the main points of the task, information on which communication protocol is used to exchange data with the system, and photos of the initial and final state of the experiment. The image storage step is generated when the system finishes learning, and after this learning, images are captured at times when there is a change in the state of the system's operation. The student feedback stage is an algorithm that generates a path of correct green circles and red circle errors that the student developed in the experiment. The evaluation occurs when the student makes wrong points, and a trail is built to show the correct path. If the flow is incorrect, opening the image to verify this point of the study is possible. To demonstrate the application of this correction system, a didactic plant from the MPS line (Modular Production System) – Festo will be used, where this experiment aims to test failed parts, aiming to store only good parts. For the initial registration, a web application will be developed to keep the necessary information and store the registration data and learning models for the test experiment in a bank. This application will also be used to store the students' feedback, showing the flow generated at the time of the experiment and the images of errors for a later conference by the teacher, knowing that the objective of this system is not to block the teaching flow, but to improve it. A quality assessment for remote experiments carried out in laboratories allows the effective evolution of the student in practical experiments.

Keywords:

Digital twins, automatic correction, mechatronics laboratories.