Infrastructure for Partially Observable Markov Decision Processes (POMDP)
Access to Parts of the Model Description
Available Actions
Add a Policy to a POMDP Problem Description
Default Colors for Visualization in Package pomdp
Estimate the Belief for Policy Graph Nodes
Helper Functions for Gridworld MDPs
Define an MDP Problem
Functions for MDP Policies
Convert between MDPs and POMDPs
Optimal action for a belief
Plot a 2D or 3D Projection of the Belief Space
POMDP Plot Policy Graphs
Extract the Policy from a POMDP/MDP
POMDP Policy Graphs
pomdp: Infrastructure for Partially Observable Markov Decision Process...
Define a POMDP Problem
POMDP Example Files
Defining a Belief Space Projection
Reachable and Absorbing States
Calculate the Regret of a Policy
Calculate the Reward for a POMDP Solution
Round a stochastic vector or a row-stochastic matrix
Sample from the Belief Space
Simulate Trajectories in a MDP
Simulate Trajectories in a POMDP
Solve an MDP Problem
Solve a POMDP Problem using pomdp-solver
Solve a POMDP Problem using SARSOP
Transition Graph
Belief Update
Value Function
Read and write a POMDP Model to a File in POMDP Format
Provides the infrastructure to define and analyze the solutions of Partially Observable Markov Decision Process (POMDP) models. Interfaces for various exact and approximate solution algorithms are available including value iteration, point-based value iteration and SARSOP. Smallwood and Sondik (1973) <doi:10.1287/opre.21.5.1071>.