All projects

Shoulder Trainer

Built the feedback system for a shoulder reduction training device, using a laser and photoresistor array to track joint movement and score accuracy in real time.

Role
Engineering Design Project | Duke University
Timeframe
Fall 2023

Problem / Goal

Medical residents typically learn to reduce dislocated shoulders on live patients. The procedure requires applying significant force at a precise angle, and an imprecise attempt carries real risk. Our team was tasked with designing a training device that would let residents practice the mechanics of a reduction before a clinical encounter, with live feedback to help them build the skill.

What I Did

My focus was the feedback system. For the trainer to be useful, it needed to track how accurately the resident moved through the reduction, not just whether they completed it.

I designed a tracking system using a laser mounted to the base of the simulated joint and a grid of photoresistors embedded in a panel at the rear of the device. To score a successful reduction, the user had to rotate the joint along the correct path, hitting each photoresistor target in sequence. A limit switch at the top of the joint registered completion, and the system returned a score based on targets hit and time taken.

I built the electronics on Arduino, starting with breadboard prototypes and finishing with a soldered shield for the final version.

Arduino board and breadboard prototype on the floor of the device frame, with wires running up to photoresistors taped to the white backing panel
Early breadboard prototype — the Arduino and photoresistor circuit before moving to a soldered shield.
Side view of the full shoulder joint assembly showing the 3D-printed humeral head sphere, glenoid socket, and PVC pipe arm extending downward, with a limit switch mounted at the joint
The joint assembly. The humeral head (lower sphere) rotates against the glenoid socket; the limit switch at the joint registers when the reduction is complete.

Results

The device received sign-off from our clinical client, Dr. Josh Broder at Duke University Hospital.

A person in a Duke Blue Devils shirt rotating the device arm upward, with the red laser dot visible near the top of the white photoresistor panel
Starting position — the laser dot sits near the top of the panel as the user begins the reduction motion.
Same view with the arm angled outward and the laser dot tracking to the middle of the panel
Mid-reduction — the arm is angled outward and the laser has moved down the panel.
Arm rotated further downward, with the laser dot near the bottom of the photoresistor panel
Final stage — the arm is lowered and the laser approaches the bottom photoresistor targets.

What I Learned

This was my first time integrating sensors and firmware into a physical system someone would actually use. Iterating on the photoresistor array taught me that real-world sensing is messier than breadboard demos suggest. Designing for a healthcare user also added a new constraint: feedback had to be immediate and unambiguous, or the trainer wouldn’t serve its purpose.