Making path planning readable without a robot in the room
An interactive robotics planning sandbox for comparing A*, Dijkstra, and an RRT-style planner with visible tradeoffs instead of a black-box animation.
Robotics work is hard to show online because the best demos often depend on hardware, lab space, cameras, maps, and setup.
I wanted a planning project that could be understood in a browser.
So I built a 2D path-planning simulator where you can paint obstacles, set a start and goal, choose an algorithm, and see the result instantly.
Why these three algorithms
A* is the practical one for grid planning when you have a good heuristic. It finds optimal paths quickly when the heuristic is admissible and does not lie about distance to goal.
Dijkstra is the baseline. It explores more because it does not use a heuristic, but that extra exploration makes it useful as a reference: you can see exactly where A* is cutting corners versus where Dijkstra is doing unnecessary work.
The RRT-style planner shows a different family of thinking — sampling through space, accepting a suboptimal result, and sometimes failing within the chosen budget. That failure behavior is important to show because it is honest about what sampling-based planners actually do.
Seeing all three side by side is more useful than reading that they exist.
What it does not pretend
This is not a full field robotics stack.
A real robot would need footprint inflation, continuous collision checking, kinematic constraints, dynamic obstacle handling, localization uncertainty, and map updates from sensors. The 2D grid is a useful abstraction for comparing algorithm behavior, not a claim about production deployment.
I like stating that plainly because the gap between demo and production is where most of the interesting work lives.