ShieldAI

Inspiration

What if something as simple as clicking a button became a daily challenge?

For people with motor disabilities or involuntary movements, navigating a computer isn’t always easy. Even head-controlled cursors can struggle with unwanted movements, making precise clicking frustrating.

That’s what inspired TremorShield AI.

We wanted to build technology that adapts to how people move instead of expecting them to move perfectly.

What it does

TremorShield AI is a webcam-based, hands-free cursor control system designed to make computer navigation steadier and more accessible.

Using AI-powered facial tracking, it translates head movements into cursor movements while applying adaptive filtering to reduce unwanted jitter.

Key features include:

Real-time head tracking using a standard webcam.
Adaptive cursor stabilization based on movement patterns.
Hands-free clicking through dwell selection.
Calibration for personalized cursor movement.
Comparison between raw tracking, traditional smoothing, and TremorShield’s adaptive filtering.

Our goal is simple: Steadier Control. Greater Independence.

How we built it

We built TremorShield AI using HTML, CSS, and JavaScript, with Google MediaPipe Face Landmarker for real-time facial tracking.

Our system follows four steps:

Detect → Map → Analyze → Adapt

Detect: Identify facial landmarks through the webcam.
Map: Convert head movements into cursor coordinates.
Analyze: Examine movement patterns to distinguish repetitive oscillations from intentional movement.
Adapt: Adjust cursor smoothing to improve stability while maintaining responsiveness.

We implemented One Euro filtering alongside our custom adaptive smoothing logic.

The prototype is deployed on Netlify, with webcam processing performed locally in the browser. We also created an interactive target-selection challenge to demonstrate and compare cursor performance.

Challenges we ran into

One of our biggest challenges was balancing cursor stability with responsiveness. Too much smoothing makes the cursor slow, while too little allows unwanted jitter.

We also faced several technical challenges:

Integrating MediaPipe and loading the facial tracking model correctly.
Fixing cursor calibration and positioning issues.
Correcting reversed horizontal cursor movement.
Maintaining smooth real-time performance.
Designing an interface that supports hands-free interaction.

Each challenge pushed us to improve both the technical implementation and the user experience.

Accomplishments that we’re proud of

We’re especially proud that we turned an accessibility-focused idea into a working, webcam-powered prototype.

Our key accomplishments include:

Building hands-free cursor navigation without specialized hardware.
Implementing adaptive motion filtering rather than relying only on basic smoothing.
Creating three tracking modes for side-by-side testing.
Developing an interactive dwell-clicking challenge with performance metrics.
Deploying a functioning browser-based prototype that others can try.

Most importantly, we built something that demonstrates how affordable technology could make digital interactions more inclusive.

What we learned

This project taught us that accessibility isn’t just about making technology usable. It’s about making it reliable, comfortable, and adaptable.

We gained hands-on experience with computer vision, facial landmark detection, real-time signal processing, cursor calibration, and browser-based development.

We also learned that combining existing AI models with thoughtfully designed algorithms can create promising solutions without training an entirely new model.

And perhaps our biggest takeaway: The best technology adapts to people, not the other way around.

What’s next for ShieldAI

Our current prototype is just the beginning.

Next, we plan to:

Build a browser extension that enables head-controlled navigation across websites.
Improve personalization with adjustable sensitivity, calibration, and dwell timing.
Conduct real-world testing with people who experience motor-control challenges.
Enhance adaptive filtering using feedback and measured performance data.
Explore broader accessibility integrations to support everyday computer interactions.

Our long-term vision is to make hands-free computer navigation affordable, accessible, and practical for everyone who needs it.

TremorShield AI — Technology That Moves With You.