Monitor Slide fit :Auto-Adjusting

What it does

Monitor Slide Fit is an intelligent system that uses facial recognition to adjust monitor distance (X), height (Z), and panning angle. It mitigates posture issues, eye strain, and manual adjustment burden, while supporting auto-reset and pause functions.

Your inspiration

This idea stems from health issues caused by prolonged monitor use, such as poor posture, muscle strain, and vision decline. Many users avoid adjusting monitors due to inconvenience or weight, leading to fixed, harmful postures. Based on real feedback, this system offers a smart monitor arm that auto-adjusts using facial recognition. Unlike manual or preset systems, it responds in real time using regression-based estimation, EMA filtering, and dead-zone logic. It solves not just convenience but also health and productivity—technically feasible and socially relevant.

How it works

This system is a smart monitor arm that automatically adjusts screen distance (X), height (Z), and angle (panning) using facial recognition. A webcam captures video, and a Raspberry Pi extracts facial center and width to estimate position via regression and relative coordinates. Position data is filtered using an EMA algorithm and dead zone logic to avoid unnecessary movement. Only significant changes generate commands, which are sent to an Arduino controlling each axis: stepper motor with ball screw (X), linear actuator (Z), and servo motor with gear reduction (panning). Axes include physical and software limits. If no face is detected for 3 seconds, the system resets to the initial position using internal pulse tracking. Gear slack is corrected via backlash compensation. The result is a stable, responsive, and precise system that adjusts monitor position in real time.

Design process

This smart monitor arm automatically adjusts screen distance (X), height (Z), and panning angle based on facial recognition. It was developed to reduce fatigue and posture-related strain by responding to user movement in real time. Originally designed as a 6-DOF robotic arm, the structure was simplified to a horizontal–vertical slide system for better control and load stability. Initial ToF and webcam inputs were streamlined to a webcam-only setup, using facial width and center coordinates to estimate distance and height. EMA filtering and dead zone logic ensure stable control. The X-axis uses a stepper motor with ball screw; the Z-axis uses a linear actuator. Limit switches and software bounds prevent overtravel. A panning function was later added using an MG996R servo, reduction gearing (2:1), and backlash compensation. If no face is detected for 3 seconds, the system resets using internal pulse tracking. Structurally, placing the X-axis above the Z-axis lowers the center of gravity, allowing the actuator to directly support the load—reducing vibration and improving rigidity. The prototype passed all tests, with FEM analysis showing a minimum safety factor of 3.78. Future improvements include metal inserts, closed-loop motors, and machine learning-based position correction.

How it is different

This project presents a smart monitor arm that automatically adjusts screen distance (X), height (Z), and panning angle using facial recognition. Unlike conventional manual or preset systems, it responds in real time to user posture for a personalized experience. Key features include regression-based distance estimation, EMA filtering, and dead zone logic, ensuring precise, stable control while minimizing noise and overreaction. Panning is enabled via a reduction gear and backlash compensation, allowing accurate head-tracking rotation beyond basic servo control. If no face is detected for 3 seconds, all axes reset to initial positions using internal pulse tracking, eliminating the need for absolute sensors and simplifying the structure. This system integrates sensing, control, and mechanics into a compact, adaptive platform—redefining ergonomic interaction with intelligent motion.

Future plans

Precision Control Replace MG996R with high-torque or closed-loop motors for improved panning accuracy and stable backlash compensation. Optimized Vision Reduce processing load via frame skipping, ROI limitation, lightweight models (e.g. MobileNet-SSD), and event-triggered inference. User-Specific Tuning Apply machine learning to adapt to user’s face, posture, and habits for personalized positioning accuracy. System Integration Compact PCB, internal cabling, and slim framing enhance usability. FEM shows safety factor >3.78, enabling dual-monitor support and scalable design.

Awards