1
Grid-Based Path Planning with A*
Model a warehouse as a discrete grid, implement A* search with an admissible Manhattan heuristic, and reason about why A* beats Dijkstra on shortest-path problems.
Explore
4.6 from 20 ratings
Amazon Robotics - Black-Friday Fleet Coordination
ProoV · Beginner - Intermediate
About this project
Step into the seat of a Robotics Operations Engineer during Black Friday: coordinate a 200+ drive-unit fleet on a warehouse grid using A* search, reservation-table cooperative planning, and surge-aware charging policies. A Proov-owned educational case study inspired by Amazon Robotics (formerly Kiva Systems) - not affiliated with, endorsed by, or sponsored by Amazon.com Inc.
Ideal for entry-level careers in
Add this to your LinkedIn profile
Projects
inAmazon Robotics - Black-Friday Fleet Coordination
Just completed
Associated with ProoV
A graded ProoV project, completed step by step and marked against a published rubric.
❖robotics, multi-agent-path-finding and more skills
What you'll do
1
Onboarding: Black Friday at Fulfillment Center XYZ
Take the seat at the traffic-control desk. Meet your floor — storage aisles, pick stations, packing lines, charging bays — and the three numbers that define your day: orders shipped, collisions, and sustained peak throughput. Watch a 20-robot cascade deadlock that frames the whole problem.
2
Phase 1: Model the Warehouse as a Grid
Reduce the warehouse to a 2D grid of typed cells, encode robot state as positions + goals + batteries, and lock in the collision definition (same cell OR same edge at the same tick). Drag five drive units onto the grid in a hands-on sandbox to build the JSON your planner will reason on.
3
Phase 2: A* for One Robot
Implement A* search in an embedded Pyodide cell — f = g + h with a Manhattan heuristic — then run your code on three procedurally generated 25×25 mazes. Reason about heuristic admissibility on a 4-connected grid.
4
Phase 3: Coordinate the Fleet
Build a centralized reservation table (cell, tick) with both vertex AND directed-edge reservations to stop head-on swaps — cooperative A*. Explore the concave throughput-vs-fleet-size curve in a live slider sandbox and find the operating point for the floor.
5
Phase 4: Tune Charging & Idle Policy
Avoid the Black-Friday thundering herd at the docks with staggered jittered charging thresholds, and disperse idle robots to perimeter cells weighted by anticipated demand — the two policy choices that let your fleet hold peak throughput through the surge.
6
Capstone: Run the Shift & Pitch the VP
Lock in your full policy on the 5-dropdown design workspace, replay an 8-hour 200-unit shift against the naive baseline in the live simulator, and defend the design in a one-page memo to the VP of Operations covering problem, evidence, engine, safety, and a scoped single-zone pilot ask.
What you'll learn
2
Multi-Agent Coordination
Scale from one robot to a fleet with a centralized reservation table indexed by (cell, tick), including vertex AND directed-edge reservations that stop head-on swaps and cascade deadlocks.
3
Operations Engineering at Surge Scale
Tune charging and idle policies to hold peak throughput through Black-Friday-scale load, read the concave throughput-vs-fleet-size curve, and defend the design in a one-page executive memo.
What students say
I gained hands-on experience in multi-agent path planning, A search, cooperative robot coordination using reservation tables, throughput optimization, charging policies, and communicating engineering decisions through an executive memo. The project helped me understand how large…