Skip to content
NOVA PYRA

Engineering Process

How we engineer

Every decision, including the ones we reject, is documented with its reasoning — a traceable record, and how new members learn why the robot looks the way it does.

The Cycle

Engineering design process

  1. 01

    Ask

    Identify the need and constraints. Used our COB strategy at the start of the season to define priorities and guide game strategy decisions.

  2. 02

    Imagine

    Investigate the problem. Analyzed risks, constraints, and gameplay challenges to inform design decisions.

  3. 03

    Make a Plan

    Select a promising solution. Developed concepts using CAD and hand-drawn designs to evaluate potential solutions.

  4. 04

    Create

    Build a prototype. Built iterative prototypes beginning with cardboard, progressing to 3D-printed parts, and refining into final designs.

  5. 05

    Test

    Evaluate the prototype. Tested prototypes in game-like conditions to measure consistency, efficiency, and performance.

  6. 06

    Improve

    Redesign as needed. Refined designs through repeated iterations, using testing data to drive continuous improvement.

Having a detailed process allows us to sustain effective prototyping and testing periods. Each year we start with a COB strategy session — identifying what is Critical, Optional, and Bypass — to create a game strategy before touching any tools. Every decision, including ones we reject, is documented in the engineering notebook with the reasoning, giving us a traceable record and helping new members understand why the robot looks the way it does.

Engineering PortfolioLive document · opens the PDF

Problem → Solution

What went wrong, and what we did

The failures are the interesting part. Each of these cost us matches before it cost us a redesign.

Problem

Intake was able to carry four balls into the intake without driver awareness, creating risk of penalties due to holding excess artifacts.

Solution

A rack and pinion feed stop acts as a hard stop for the first incoming artifact, defining a fixed position within the intake. This limits available space for additional artifacts, preventing a fourth artifact from being collected.

Result

Penalty risk eliminated — intake is mechanically prevented from collecting a fourth artifact.

Problem

At higher flywheel speeds, artifacts gained excessive backspin along the hood interface, resulting in bounce-outs, reduced consistency, and less than 75% shot accuracy.

Solution

Tested multiple hood and flywheel interactions, iterating on compression and material contact to reduce backspin.

Result

Shot consistency improved above 75% accuracy through iterative hood and flywheel compression design.

Problem

To reduce backspin, we added kicker wheels — but experienced inconsistencies in tuning and shot performance.

Solution

Sometimes newer isn't better. After testing, we determined that added complexity reduced reliability. We removed the kicker wheels and refined our original design.

Result

Returning to the refined original design achieved more consistent shot performance than the kicker wheel approach.

Problem

Without proper tuning, shot speed varied between cycles, leading to inconsistent scoring and reduced reliability.

Solution

Implemented RPM-based PID control to maintain consistent flywheel speed, stabilizing shot performance and allowing quick adjustments between shots.

Result

Flywheel speed stabilized — reduced shot variability and consistent scoring across all match phases.

Testing & Data

We measured it

Every claim about the robot traces back to a number we recorded.

Shot Accuracy by Development Stage

% of shots on target (50+ repetitions per stage from fixed field position)

↑ Higher is better

  • Initial Flywheel74%
  • Hood Iteration83%
  • Removed Kicker Wheels88%
  • RPM PID Tuned95%

> RPM PID flywheel control lifted accuracy from 74% → 95% across four build iterations

Driver Inputs Required per Scoring Cycle

Button presses needed to complete one full intake-to-shoot cycle

↓ Lower is better

  • Manual Control6 inputs
  • Partial Automation3 inputs
  • State Machine (Final)1 inputs

> State machine automation reduced driver cognitive load by 83% — one button completes the full cycle

Shot Consistency — Lookup Table Refinement

Standard deviation of shot landing position (inches) across 50 trials

↓ Lower is better

  • Pre-Lookup Table8"
  • v1 Lookup Table5"
  • v2 + PID2"
  • Final (Sensor FB)1"

> Sensor feedback closed the loop — shot spread tightened from 8" to 1" standard deviation