Project Gallery
01 Project Overview
Three AGVs, one crowded warehouse, six deliveries — and zero collisions. This MATLAB simulation, built for the Robot Path Planning and Mobile Robots course at Manipal Institute of Technology, pairs a cooperative A* planner that predicts and dodges other robots' paths with a reactive controller that guarantees safety in real time. The real payoff is the analysis: it doesn't just prove the fleet works, it hunts down and explains the one robot whose planner made an oddly cautious call.
02 Key Features & Achievements
3-robot AGV fleet cooperatively completing 6 tasks on a 40x40 warehouse grid
Modified A* planner treats other robots' predicted paths as dynamic soft obstacles
Reactive emergency-stop layer guarantees zero real collisions
Path smoothing removes redundant A* waypoints for efficient trajectories
Custom analytics pipeline scoring speed utilization and path-following accuracy
03 Technical Stack
04 Challenges & Solutions
Challenge 1
Keeping three robots collision-free on a shared, congested grid
Solution
Layered a predictive A* cost model with a reactive emergency-stop controller, achieving zero true collisions.
Challenge 2
Robot 2 planned a path nearly double its actual length (207 vs. 109 cells)
Solution
Traced it to an oversized conflict-penalty weight in the A* cost function, pinpointing the exact fix needed.
05 Key Achievements
Zero collisions across all six delivery tasks
95.14% fleet-average speed utilization
92.66% and 89.14% path-following accuracy for Robots 1 and 3
Diagnosed a planning anomaly in Robot 2 (52.61% accuracy) to a specific over-penalized cost term