Overview
What this challenge is about.
You receive a discrete-event simulator of a 1,200-shelf warehouse with calibrated optical-scanning error rates and stock-out cost per shelf. Formulate the restocking decision as a POMDP and implement a planner (POMCP or a smaller belief-MDP approximation). Compare against the greedy heuristic across 100 simulated 8-hour shifts on: stock-out incidents, robot idle time, and total restocking cost. The memo recommends keep, swap, or staged rollout.
The Brief
What you'll do, and what you'll demonstrate.
Build a POMDP planner that reduces stock-outs in a real warehouse simulator without inflating robot idle time.
Earning criteria — what you'll demonstrate
- Formulate a real-world problem as a POMDP
- Implement and tune a Monte Carlo planner over beliefs
- Compare planners on operations-level KPIs, not just expected reward
- Communicate planner behavior to an operations audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
AI Engineer
Shipping a planner against real operational KPIs and writing an ops-team rollout plan is the AI-engineer pattern at warehouse-tech companies.
This challenge sharpens
- pomdp
- planning-under-uncertainty
- simulation
ML Researcher
Sensitivity-analysis discipline and faithful problem formulation are the rigor markers ML researchers carry across applied projects.
This challenge sharpens
- monte-carlo-planning
- simulation
- operations-research