Browsing by Author "supervised by dr : KhelifiRamzi"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item RotativeRoboticArmforIntelligentFace Tracking(المركز الجامعي إليزي - معهد العلوم، قسم الإعلام الآلي, 2026-09-23) TagabouDoua; supervised by dr : KhelifiRamziTeleGimbalisanintelligent rotatingroboticarm powered byartificial intelligenceforfacetracking. Thesystemaimstosolvetheproblemoftargetlosscausedbythecontinuousmovementofspeakers inside lecture halls, where fixed cameras or human operators often fail to track movement reliably and cost-effectively. The system relies on computer vision algorithms using the OpenCV library, where faces are detected using the YuNet model and recognized using the SFace model. Based on the visual coordinatesobtainedfromimageprocessing,thesystemsendsreal-timecontrolcommandsthrough an Arduino UNO microcontroller to drive the Pan rotation axis and the Tilt axis using NEMA 17 stepper motors controlled by A4988 drivers with 1/16 microstepping enabled, providing smooth, silent, and vibration-free motion. Five major engineering challenges were documented and addressed: audio resonance, solved through microstepping activation; overshoot caused by artificial intelligence processing, solved by applyinga50-pixeltolerancethresholdandlimitingspeed;mechanicalvibration,solvedbyphysical adjustmentsand increasingthemountingheight;target loss,solvedthrough asafetymechanism that immediately sends a stop command; and stepper motor overheating, solved by adjusting the VREF value to 0.8 volts. Experimentalresultsdemonstratedhighoperationalefficiency,achievingframeratesrangingfrom 27.75to29.79framespersecond,face recognitionconfidenceexceeding0.7,responsetimesbelow 50 milliseconds, and an angle measurement accuracy reaching 92.14 degrees. From a commercial perspective, the feasibility study showed a gross profit margin of 71.6%, a break-even point at 41 units sold, and a return on investment reaching 133.6% at 200 units sold. ThisacademicprototypewasdevelopedwithintheframeworkofMinisterialDecisionNo.1275and is proposed as a marketable commercial product under the startup project TeleGimbal.