Automated Warehouse Planning System

A logic-based planning system for coordinating autonomous warehouse robots on a grid, built using Answer Set Programming (ASP) with Clingo — Master's coursework project at ASU.

The scenario models an Amazon-style warehouse floor as a grid, like a chessboard. Robots — think Roomba-like platforms carrying shelves on top — need to navigate from starting positions to destinations without colliding with each other.

I defined the grid, its bounds, and the robots as logical entities with explicit capabilities and constraints using ASP. The solver's job was to find valid sequences of moves that get every robot to its goal. Clingo would enumerate answer sets — each one a complete, valid plan of coordinated movements across all robots.

The constraints were the real work. Robots couldn't occupy the same tile at the same time. Termination constraints had to be defined mathematically to prevent the solver from running infinite loops. And the scenarios varied in grid size, number of robots, and how tightly packed the starting configurations were.

The hardest problems came from concurrency. Early on, I allowed all robots to move simultaneously, which created deadlock situations — robots endlessly shuffling around each other in tight spaces without ever reaching their goals. The analogy is losing someone in a grocery store where you're both moving: you keep walking away from each other and never converge. The fix required rethinking how simultaneous movement was constrained so that the solver could find valid orderings even in cramped configurations.

The solution handled all provided problem instances, though the larger and more constrained scenarios took significantly longer to solve — some running over a full day. That scalability wall is inherent to ASP on combinatorially complex problems, and understanding where and why it hits was one of the more valuable takeaways from the project.