Abstract:Partially observable Markov decision processes (POMDPs) provide a flexible representation for real-world decision and control problems. However, POMDPs are notoriously difficult to solve, especially when the state and observation spaces are continuous or hybrid, which is often the case for physical systems. While recent online sampling-based POMDP algorithms that plan with observation likelihood weighting have shown practical effectiveness, a general theory characterizing the approximation error of the particle filtering techniques that these algorithms use has not previously been proposed. Our main contribution is bounding the error between any POMDP and its corresponding finite sample particle belief MDP (PB-MDP) approximation. This fundamental bridge between PB-MDPs and POMDPs allows us to adapt any sampling-based MDP algorithm to a POMDP by solving the corresponding particle belief MDP, thereby extending the convergence guarantees of the MDP algorithm to the POMDP. Practically, this is implemented by using the particle filter belief transition model as the generative model for the MDP solver. While this requires access to the observation density model from the POMDP, it only increases the transition sampling complexity of the MDP solver by a factor of $\mathcal{O}(C)$, where $C$ is the number of particles. Thus, when combined with sparse sampling MDP algorithms, this approach can yield algorithms for POMDPs that have no direct theoretical dependence on the size of the state and observation spaces. In addition to our theoretical contribution, we perform five numerical experiments on benchmark POMDPs to demonstrate that a simple MDP algorithm adapted using PB-MDP approximation, Sparse-PFT, achieves performance competitive with other leading continuous observation POMDP solvers.

SHPOS: A Theoretical Guaranteed Accelerated Particle Optimization Sampling Method

An Improved APSO-SQP with Adaptive Transition Strategy for Dynamic Optimization

Enhancing Particle Swarm Optimization via probabilistic models

A Method of Parameter Optimization for Particle Swarm Optimization Based on Stochastic Processes

Dynamic Heterogeneous Particle Swarm Optimization.

Sample Particle Swam Optimization and Its Dynamic Behavior

A Unified Particle-Optimization Framework for Scalable Bayesian Sampling

A stochastic scattering particle swarm optimizer

Stochastic Momentum Method with Double Acceleration for Regularized Empirical Risk Minimization

A Quantum-Based Particle Swarm Optimization Algorithm Applied to Inverse Problems

Particle Optimization in Stochastic Gradient MCMC

Parallel Simultaneous Perturbation Optimization

Faster Sampling via Stochastic Gradient Proximal Sampler

OPES: On-the-fly Probability Enhanced Sampling Method

Optimality Guarantees for Particle Belief Approximation of POMDPs

A Splitting Hamiltonian Monte Carlo Method for Efficient Sampling

cuPSO: GPU Parallelization for Particle Swarm Optimization Algorithms

Convergence and Stability of the Stochastic Proximal Point Algorithm with Momentum

Proximal Policy Optimization Smoothed Algorithm

Efficient Optimization Procedures for Stochastic Simulation Systems

Sharper Analysis for Minibatch Stochastic Proximal Point Methods: Stability, Smoothness, and Deviation.