// the find
ai-winter/matlab_motion_planning
Motion planning and Navigation of AGV/AMR：matlab implementation of Dijkstra, A*, Theta*, JPS, D*, LPA*, D* Lite, RRT, RRT*, RRT-Connect, Informed RRT*, ACO, Voronoi, PID, LQR, MPC, APF, RPP, DWA, DDPG, Bezier, B-spline, Dubins, Reeds-Shepp etc.
A MATLAB collection of classic motion planning algorithms for AGVs/mobile robots — graph search (Dijkstra, A*, JPS, Theta*, D*), sampling-based planners (RRT family), local planners (DWA, MPC, APF, PID, LQR), and curve generation (Bezier, Dubins, Reeds-Shepp). It's for robotics students and researchers who want a reference implementation to read, not a library to drop into a product.
Breadth is real — one repo covers search-based, sampling-based, evolutionary, and local planning with consistent visualization (plot_grid, plot_expand, plot_path) across all of them, so you can compare algorithms apples-to-apples on the same grid maps. Every algorithm links back to the actual paper it implements, which is rare and useful if you're trying to understand the theory rather than just call a function. The .mlx live-script examples (simulation_global.mlx, simulation_local.mlx) make it easy to swap planner_name and immediately see animated output without touching plotting code.
It's MATLAB, so there's a licensing wall before you can even run it, and no way to use this in a real robot stack without a full reimplementation — the ROS C++ and Python sister repos are separate, maintained-by-the-same-author projects, not this one. No tests of any kind, so 'done-v1.0' badges are just the author's word that the code runs once; there's no way to verify correctness or catch regressions. Several entries (LPA*, D* Lite, TEB, GA, PSO, ABC) are marked 'develop' with no animation and have been in that state for a while going by the README — the badge matrix overstates what's actually finished. No API docs or function signatures beyond reading the .m files directly; everything is discovered by opening source.