EdgeSense

Inspiration

Our team built EdgeSense to address a critical challenge in responsible AI improving workplace safety without compromising employee privacy. Traditional workplace monitoring can introduce invasive surveillance and data security concerns. Inspired by IBM's emphasis on ethical, transparent, and human centered AI, we wanted to demonstrate that worker protection and individual autonomy can coexist

What it does

EdgeSense is a privacy first ergonomic safety assistant that uses edge based computer vision and skeletal pose estimation to detect potentially unsafe lifting techniques, awkward posture, and unbalanced grips. It provides real time corrective recommendations and explains the detection evidence behind them while minimizing the transmission of identifiable data.

How we built it

Our team developed EdgeSense using Next.js, React, Typescript, Python, FastAPI, and pose estimation technology. Our system combines local video processing, skeletal coordinates, ergonomic risk detection, and explainable recommendations. We also used IBM Bob as an agentic AI Development partner to help develop and test our recommendation explanation feature, strengthening transparency and accountability.

Challenges we ran into

The biggest challenges we ran into was maintaining worker Privacy while interpreting physical movements accurately, developing convincing 3D skeletal visualizations, and integrating independently developed frontend and backend systems. We also had to account for the limitations of single camera pose estimation

Accomplishments that we're proud of

We built a functioning live ergonomic assessment prototype with privacy-conscious skeletal visualization, realtime risk detection, and corrective action guidance. Our IBM Bob assisted explanation feature shows users why a recommendation was selected and acknowledges the limitations of the underlying detection

What we learned

We learned that Responsible AI in business requires more than technical performance, Transparency, data minimization, human oversight, and honest communication of system limitations are equally important when build AI that people can trust.

What's next for EdgeSense

We plan to improve our detection accuracy, expand our ergonomic risk recognition, validate our recommendations against workplace safety standards, and explore scalable privacy preserving analytics that help organizations prevent injuries without relying on invasive employee surveillance