Research Output
Adaptive PD Controller Performance for Direct Cooling of Thermoelectric Refrigerator
Real-time vision-based hand gesture to text interpreter by using artificial intelligence with augmented reality element
1 - 2 of 2
Publication
Refrigerator is the key component to keep the medicine and biological sample in the hospital. The domestic refrigerator has the problem of larger size and heavier weight since to the compact system like condenser, compressor, evaporator and expansion valve are assemble and using in the refrigerator. This project focused on design of temperature control of the portable thermoelectric refrigerator for medical purpose. Thermoelectric refrigerator is using the direct cooling method through thermoelectric module. Thermoelectric refrigerator has several advantages such as smaller size, lighter and silent when operated. Since maintain a constant temperature for the storage of medical product is important, a specific refrigerator is needed to ensure the medicine is stored in desired temperature. This project is to design and develop an adaptive control system which can perform a good temperature control for the thermoelectric refrigerator. The second order model is applied to design adaptive Proportional-Derivative (PD) controller. The selected controller is the adaptive PD controller because the performance of response shows 0.42 C of less steady state error and 0.21 C of lower undershoot. The adaptive PD control system designed able to let the refrigerator operate in different operating condition without influence the performance of the refrigerator.
Publication
Real-time Vision-based Hand Gesture to Text Interpreter by Using Artificial Intelligence with Augmented Reality Element is a device that can interpret sign language to text in real-time. This communicator used a machine learning approach with a slight touch of deep learning elements, which are OpenCV, MediaPipe, and TensorFlow algorithms. Those algorithms have been used to differentiate the hand from other objects, detect the movement and coordinate of hands and perform imagery data analysis to produce output instantly in real-time. The camera will detect the user's hand movement, and the output will be produced on an LCD monitor. This project has been developed by using Python programming language. 13,000 of ASL's alphabets and 5,000 of ASL's number imagery datasets have been collected and trained by using cloud platforms which are Google Teachable Machine and Google Colab. The training process produced 99.85% of accuracy for the alphabets and 100% accuracy for the number. Finally, the constructed machine learning algorithm able to display alphabets and numbers on an LCD monitor by performing ASL's alphabet and number hand gesture in real-time. The performance of the prototype has been analyzed and experimented by two users at plain and noise background with different determined distances.