Skip to content

Introduction

Welcome, explorers, to the vibrant realm of “PacBot” one of the thrilling themes of eYRC 2026-27! This year we step into a maze where a lone bot must gather every pellet it can and slip past the Ghosts on its way out.


Imagine a grid of narrow corridors and blind corners, Pellets scattered across it, and a single exit waiting on the other side. The full layout is known from the start but knowing the map and surviving it are two very different things.

Deep inside the maze a lone bot powers up. Somewhere in the same corridors a Ghost bot is already patrolling: pre-programmed, relentless, and closing in. Every corner your bot turns, the Ghost is turning one too.

Your mission is to bring this escape to life in simulation first, then on real hardware:

  • Plan a route that collects the maximum points from the pellets scattered across the grid
  • Track the Ghosts and reroute in real time as their positions change every second
  • Reach the exit before the escape routes are cut off

It’s about the trade-off between greed and survival: every detour for one more pellet is time the Ghost spends getting closer.


DomainSkills
Maze Solving & Path PlanningRepresenting the maze as a graph, running search algorithms over it, scoring a route by the points it wins against the time it costs
Strategy Under PressureDeciding how greedy to be, tracking where the Ghosts are, replanning the instant a corridor stops being safe
Motion & ControlDriving corridors accurately, clean turns at junctions, reaching the exit inside the time limit
Simulation & MessagingWorking with the MuJoCo model of the maze and bot, and MQTT as the link between your code and the robot


This is your starting point. Everything you build across both stages — the planner, the Ghost tracking, the escape run — traces back to the mission laid out here.

Head over to Task 0