EDUCATIONAL ROBOTIC GRIPPER CONTROLLED BY COMPUTER VISION (MEDIAPIPE) AND ARDUINO HARDWARE WITH MANUAL REDUNDANCY
The integration of Digital Information and Communication Technologies (DICT), Educational Robotics, and Artificial Intelligence (AI) into secondary and technical education faces the ongoing challenge of bridging abstract mathematical and physical concepts with hands-on laboratory experimentation[cite: 3]. This paper presents the development, cyber-physical architecture, and experimental validation of a low-cost Educational Robotic Gripper built upon the Arduino platform and a Python desktop application (Tkinter, OpenCV, and MediaPipe)[cite: 3]. The system provides redundant bimodal control: contactless remote actuation via real-time gesture tracking of 21 three-dimensional hand landmarks through a standard webcam, alongside local manual operation via an analog joystick module and push-button[cite: 3]. To prevent concurrency conflicts between both control layers, a non-blocking temporal arbitration policy based on the millis() function was implemented in the embedded firmware, granting a 1,000 ms priority lock to serial commands issued by the computer vision interface[cite: 3]. Bench validation demonstrated fluid gesture-to-actuator responsiveness, mechanical stability using an SG90 micro-servo, and seamless switching between autonomous and manual modes[cite: 3]. The resulting cyber-physical artifact offers a replicable, interdisciplinary teaching tool for affine functions, coordinate geometry, kinematics, basic electronics, and computational thinking[cite: 3].
Authors
- Amarildo Ferreira Oliveira
- Danielle da Silva Salvaterra
- Vlademir de Paula Gabriel
- Elenilson Rodrigues
Institutions
- Instituto Federal de Educação Ciência e Tecnologia do Mato Grosso do Sul (BR)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24373615
- Primary Topic
- Educational Robotics and Engineering
- Type
- article
- Field-Weighted Citation Impact
- 0.00