PortfolioProjectView

PrismHighlight

Button

SupabaseUserGlobalContext

VisitorTracker

Screen

UserLoginState

[Library] Tailwind 3.4 Number Tokens

react-awesome-reveal

[Library] Tailwind 3.4 Color Tokens

lottie-react

vanilla-cookieconsent

date-fns

plasmic-nav

plasmic-embed-css

loading-boundary

plasmic-query

uuid

copy-to-clipboard

plasmic-basic-components

[Library] Tabler 3.2 Icons

Low Inertia

react-youtube

plasmic-tabs

react-aria

plasmic-cms

react-twitter-widgets

plasmic-rich-components

react-chartjs-2

antd5 hostless

tiptap

react-slick

react-parallax-tilt

react-quill

maze-solving-robot-with-fastest-return-logic

<p data-start="165" data-end="382"><strong data-start="165" data-end="183">Project Title:</strong> Maze Solver with Fastest Return Logic (LEGO NXT 3.0, C)<br data-start="239" data-end="242">
<strong data-start="242" data-end="255">Timeline:</strong> 1 week<br data-start="262" data-end="265">
<strong data-start="265" data-end="274">Team:</strong> Team project (shared responsibilities)<br data-start="313" data-end="316">
<strong data-start="316" data-end="329">My focus:</strong> Hardware compacting + PID heading tuning &amp; testing</p>
<h2 data-start="384" data-end="542"><strong data-start="384" data-end="401">What it does:</strong></h2><p data-start="384" data-end="542">A LEGO NXT robot that solves a tiled maze from a start point, reaches the end, then returns to the start using an optimized fastest route.</p>
<h2 data-start="544" data-end="654"><strong data-start="544" data-end="553">Goal:</strong></h2><p data-start="544" data-end="654">Solve + return in <strong data-start="574" data-end="589">&lt; 2 minutes</strong> with <strong data-start="595" data-end="614">no wall contact</strong> and <strong data-start="619" data-end="653">no unnecessary moves on return</strong>.</p>
<h2 data-start="656" data-end="674"><strong data-start="656" data-end="672">Performance:</strong>
</h2><p data-start="677" data-end="704">Fastest run: 1:55<strong data-start="707" data-end="724"></strong><strong data-start="707" data-end="724"></strong></p><p data-start="677" data-end="704"><strong data-start="707" data-end="724">Success rate:</strong> ~<strong data-start="726" data-end="733">80%</strong> (main failure mode: uneven ground causing heading drift)</p>

<p></p>


<h2 data-start="833" data-end="849">Requirements</h2><ul><li>No unnecessary moves on the way back</li><li>Must not touch the wall</li><li>Programmed in <strong data-start="935" data-end="940">C</strong></li><li>Full run under <strong data-start="960" data-end="973">2 minutes</strong></li></ul><h2>
Constraints</h2><ul><li>Tile size ~<strong data-start="1004" data-end="1021">30 cm × 30 cm</strong> (tight alignment tolerance)</li><li>Only <strong data-start="1059" data-end="1082">one sensor allowed:</strong> directional ultrasonic</li><li>Only LEGO NXT parts</li><li>Limited compute/motor precision → lightweight control needed</li></ul> 
<h2 data-start="1196" data-end="1217">Solution Overview</h2>
<p data-start="1218" data-end="1247"><strong data-start="1218" data-end="1247">Navigation (Solve phase):</strong></p><p data-start="1218" data-end="1247"></p><ul><li>Used <strong data-start="1255" data-end="1284">right-hand wall-following</strong> with a single directional ultrasonic sensor.</li><li>Designed for consistent tile-to-tile movement without extra sensing </li></ul><p><strong data-start="1402" data-end="1442">Fastest Return Logic (Return phase):</strong>
</p><ul><li>Logged movement decisions during exploration.</li><li>When a dead-end was found and backtracking occurred, the robot <strong data-start="1556" data-end="1585">removed the dead-end loop</strong> from the path log, producing a shorter optimized return route.</li><li>Result: return path avoids redundant moves and is faster than retracing the full exploration path.</li></ul><p data-start="1445" data-end="1490"><br></p><h2 data-start="1445" data-end="1490">Hardware:</h2><ul><li>Built a <strong data-start="1775" data-end="1802">small-footprint chassis</strong> to prevent wall clipping.</li><li>Packed components tighter (including upright CPU positioning) to reduce width.</li></ul>







<p></p>






<h2 data-start="1975" data-end="2016">Key Engineering Decisions &amp; Tradeoffs</h2><ul><li><strong data-start="2017" data-end="2040">Speed vs. accuracy:</strong>
Higher motor speed increased overshoot and misalignment risk. Final tuning prioritized repeatability and wall clearance while still meeting the 2-minute target.</li><li><strong data-start="2205" data-end="2244">Heading stability on real surfaces:</strong>
Uneven ground caused heading drift and reduced success rate. I focused on <strong data-start="2321" data-end="2344">PID heading control</strong> tuning to smooth corrections and reduce accumulated error.</li></ul>

<h2 data-start="2405" data-end="2428">Testing &amp; Iteration</h2><ul><li>Tested in a <strong data-start="2443" data-end="2455">6×4 maze</strong></li><li>Issues found:</li></ul><ol><li><p data-start="2478" data-end="2534">Overshoot at high speed → slowed down + tuned PID values</p></li><li><p data-start="2478" data-end="2534">Larger builds clipped walls → compacted hardware layout</p></li><li><p data-start="2478" data-end="2534">Uneven ground caused heading loss → refined heading PID</p>
 </li></ol><h2 data-start="2656" data-end="2677">Improvements (V2)
</h2><ul><li>Add sensors on multiple sides (if allowed) to reduce time spent turning/scanning.</li><li>Increase chassis rigidity/traction to reduce heading drift on uneven tiles.</li><li>Further reduce footprint and improve turning precision to push success rate beyond 90–95%.</li></ul>




<p></p>


Maze Solving Robot With Fastest-Return Logic

Mechatronics engineering student with expertise in robotics, CAD design, and hardware troubleshooting. Skilled in Python, C++, and industrial tools with proven project delivery experience.