This archive contains an implementation of several optimistic planning algorithms, together with examples on how they can be used.
Author: Lucian Busoniu

Getting started
- Unzip the archive into a directory of your choice.
- Start up Matlab, point it to the directory where you unzipped the file, and run startup.
- Open the script onlineplanning_demo and try out the algorithms

Planning algorithms
- opd.m: OPD, Optimistic Planning for Deterministic Systems
- opss.m: OP-MDP, Optimistic Planning for MDPs (also called "for Sparsely Stochastic Systems")
- olop.m: OLOP, Open-Loop Optimistic Planning, a practical variant
- olop_theoretical.m: OLOP, Open-Loop Optimistic Planning, the theoretical variant given in the original paper
- sooplp.m: SOOP, SOO for Planning, and LP, Lipschitzian Planning (continuous-action algorithms)

These are usually called indirectly via the genmpc function, which deals with the online closed-loop control.

Building your own problems
The archive includes an implementation of the inverted pendulum. To build your own (simulation) problems, you need to implement two functions: one ending in "_problem" which creates the simulation model, and one ending in "_mdp" which implements the dynamics and reward function of the MDP. Descriptions of the behavior these functions must implement are found in sample_problem and sample_mdp.

Software requirements
The algorithms require Matlab 7.3 (R2006b) or later, with the Statistics toolbox included.

Contact
If you get stuck anywhere using the code, chance upon bugs or missing functions, or have any questions, comments, or suggestions, please contact the author at lucian@busoniu.net.