PiersonJones

Context

I lead a team of 4 designing a microcontroller-based object detection system to reduce scissor lift collisions. The project is developed with support from Duke's EGR 101/102 engineering design sequence, with Skanska as the industry partner and a power consultant advising on energy use.

Approach

• Generation 1 was an ultrasonic prototype: two sensors on a sweeping servo, with trigonometry on the servo angle to place objects relative to the lift and a radar visualization on a laptop. The sweep limited speed and coverage and drew heavy current, which cut battery life.
• Generation 2 replaced it with six VL53L8CX time-of-flight sensors sharing one I2C bus through a TCA9548 multiplexer. The firmware is a non-blocking C++ state machine that switches ranging between 1 Hz when nothing is near and 6 Hz when something is, with hysteresis and a 4-second hold before stepping back down. The buzzer steps one level at a time so alerts ramp smoothly.
• An 11-part test suite characterized the sensor's power draw: ranging frequency, 4x4 versus 8x8 resolution, integration time, ranging mode, ESP32 light sleep with interrupt wake, duty cycling and calibration tuning.

Result

Energy use dropped from 627 mAh to 109.2 mAh over a 10-hour run, an 83% reduction, by interfacing directly with the sensor's APIs. Next up is a KiCad PCB design to shrink the product's footprint.

Context

An independent contract project to reproduce the timing behavior of a legacy analog encoder circuit, a 555 timer oscillator, in register-driven digital logic.

Approach

• Reverse-engineered the circuit's timing and derived the register values needed to reproduce it digitally, correcting a data-entry error in the client-supplied specification along the way.
• Programmed an Arduino hardware timer (Timer1) in CTC mode with a compare-match interrupt to generate precise signal timing, replacing an analog RC oscillator.
• Diagnosed a simulator limitation across two test platforms (Tinkercad and Wokwi) and redesigned the measurement method around a single-board self-loopback interrupt.

Result

Under 2% timing error against target, versus 8 to 13% for the original analog circuit.

Context

A team of 4 built a Python simulation of the ISS deorbit, reentry, and rocket ascent, structured as four sequential phases.

Approach

4th-order Runge-Kutta integration solves the position and velocity differential equations. Gravitational and atmospheric drag forces are modeled, and event detection with interpolation pinpoints stage separation and impact.

Result

The full run simulates about 180 minutes from deorbit to splashdown and plots the trajectory and altitude profiles with NumPy and Matplotlib.

Context

A multi-robot coordination challenge. Each robot follows a track, does its own sensing task, and shares its result with the rest of the team so the group can total its results together. Mine used a Hall effect sensor to find a hidden magnet.

Approach

• Sensor-fusion steering that combines three reflectance readings into a single state driving dual continuous-rotation servos.
• A Hall effect sensor subsystem that detects a hidden magnet at a designated checkpoint and reports it through an external LED.
• Wireless communication over XBee, encoding each robot's score as a single transmitted byte using disjoint ASCII ranges per team.

Result

An LCD scoreboard, driven over SoftwareSerial, totals every team's score in real time, followed by a scripted RGB light show on the on-board and external LEDs or a note sequence through the LCD's speaker.

What I do

• Contributing to the design of two AUVs competing at the annual RoboSub competition.
• Designing an independent claw and gripper concept for the team's third, newest robot using Autodesk Fusion 360 and SolidWorks.
• Leading and onboarding 3 new members on the claw and gripper subteam, teaching CAD fundamentals and guiding each through designing their own concept for a comparative build-and-test process that picks this year's design.
• Collaborating with the ECE and CS subteams to improve component integration for the claw and gripper.
• CNC machined and fabricated buoyancy components through CNC tool-pathing in Autodesk Fusion.

Approach

• Generated all 12 equal-tempered note frequencies from A440, then derived six octaves of every pitch and its enharmonic flat algebraically.
• Synthesized each note as a fundamental plus up to two overtones under an exponential decay envelope (0.45 s time constant), adding harmonics progressively as the piece develops.
• Converted 131 BPM and standard note-length ratios, including tied notes, into discrete-time sample vectors at 44.1 kHz.
• Built the treble melody and a four-note chord bass line as independent voices, matched their lengths, and summed them into one signal.

Result

Added a delayed echo, normalized the signal, and exported a 73-second .wav file.