This paper introduces a Monte-Carlo algorithm for online planning in large POMDPs. The algorithm combines a Monte-Carlo update of the agent’s belief state with a Monte-Carlo tree search from the current belief state. The new algorithm, POMCP, has two important properties. First, Monte-Carlo sampling is used to break the curse of dimensionality both during belief state updates and during planning. Second, only a black box simulator of the POMDP is required, rather than explicit probability distributions. These properties enable POMCP to plan effectively in significantly larger POMDPs than has previously been possible. We demonstrate its effectiveness in three large POMDPs. We scale up a well-known benchmark problem, rocksample, by several orders of magnitude. We also introduce two challenging new POMDPs: 10 × 10 battleship and partially observable PacMan, with approximately 10 18 and 10 56 states respectively. Our Monte-Carlo planning algorithm achieved a high level of performance with no prior knowledge, and was also able to exploit simple domain knowledge to achieve better results with less search. POMCP is the first general purpose planner to achieve high performance in such large and unfactored POMDPs. 1

The Partially Observable Markov Decision Process (POMDP) framework has proven useful in planning domains where agents must balance actions that provide knowledge and actions that provide reward. Unfortunately, most POMDPs are complex structures with a large number of parameters. In many real-world problems, both the structure and the parameters are difficult to specify from domain knowledge alone. Recent work in Bayesian reinforcement learning has made headway in learning POMDP models; however, this work has largely focused on learning the parameters of the POMDP model. We define an infinite POMDP (iPOMDP) model that does not require knowledge of the size of the state space; instead, it assumes that the number of visited states will grow as the agent explores its world and only models visited states explicitly. We demonstrate the iPOMDP on several standard problems. 1

General purpose intelligent learning agents cycle through (complex,non-MDP) sequences of observations, actions, and rewards. On the other hand, reinforcement learning is welldeveloped for small finite state Markov Decision Processes (MDPs). So far it is an art performed by human designers to extract the right state representation out of the bare observations, i.e. to reduce the agent setup to the MDP framework. Before we can think of mechanizing this search for suitable MDPs, we need a formal objective criterion. The main contribution of this article is to develop such a criterion. I also integrate the various parts into one learning algorithm. Extensions to more realistic dynamic Bayesian networks are developed in the companion article [Hut09].

Feature Markov Decision Processes (ΦMDPs) [Hut09] are well-suited for learning agents in general environments. Nevertheless, unstructured (Φ)MDPs are limited to relatively simple environments. Structured MDPs like Dynamic Bayesian Networks (DBNs) are used for large-scale realworld problems. In this article I extend ΦMDP to ΦDBN. The primary contribution is to derive a cost criterion that allows to automatically extract the most relevant features from the environment, leading to the “best ” DBN representation. I discuss all building blocks required for a complete general learning algorithm.

www.hutter1.net General-purpose, intelligent, learning agents cycle through sequences of observations, actions, and rewards that are complex, uncertain, unknown, and non-Markovian. On the other hand, reinforcement learning is well-developed for small finite state Markov decision processes (MDPs). Up to now, extracting the right state representations out of bare observations, that is, reducing the general agent setup to the MDP framework, is an art that involves significant effort by designers. The primary goal of this work is to automate the reduction process and thereby significantly expand the scope of many existing reinforcement learning algorithms and the agents that employ them. Before we can think of mechanizing this search for suitable MDPs, we need a formal objective criterion. The main contribution of this article is to develop such a criterion. I also integrate the various parts into one learning algorithm. Extensions to more realistic dynamic Bayesian networks are developed in Part

Robots must perform tasks efficiently and reliably while acting under uncertainty. One way to achieve efficiency is to give the robot commonsense knowledge about the structure of the world. Reliable robot behaviour can be achieved by modelling the uncertainty in the world probabilistically. We present a robot system that combines these two approaches and demonstrate the improvements in efficiency and reliability that result. Our first contribution is a probabilistic relational model integrating common-sense knowledge about the world in general, with observations of a particular environment. Our second contribution is a continual planning system which is able to plan in the large problems posed by that model, by automatically switching between decision-theoretic and classical procedures. We evaluate our system on object search tasks in two different real-world indoor environments. By reasoning about the trade-offs between possible courses of action with different informational effects, and exploiting the cues and general structures of those environments, our robot is able to consistently demonstrate efficient and reliable goal-directed behaviour. 1

From an automated planning perspective the problem of practical mobile robot control in realistic environments poses many important and contrary challenges. On the one hand, the planning process must be lightweight, robust, and timely. Over the lifetime of the robot it must always respond quickly with new plans that accommodate exogenous events, changing objectives, and the underlying unpredictability of the environment. On the other hand, in order to promote efficient behaviours the planning process must perform computationally expensive reasoning about contingencies and possible revisions of subjective beliefs according to quantitatively modelled uncertainty in acting and sensing. Towards addressing these challenges, we develop a continual planning approach that switches between using a fast satisficing “classical ” planner, to decide on the overall strategy, and decision-theoretic planning to solve small abstract subproblems where deeper consideration of the sensing model is both practical, and can significantly impact overall performance. We evaluate our approach in large problems from a realistic robot exploration domain.

www.hutter1.net Specialized intelligent systems can be found everywhere: finger print, handwriting, speech, and face recognition, spam filtering, chess and other game programs, robots, et al. This decade the first presumably complete mathematical theory of artificial intelligence based on universal induction-predictiondecision-action has been proposed. This information-theoretic approach solidifies the foundations of inductive inference and artificial intelligence. Getting the foundations right usually marks a significant progress and maturing of a field. The theory provides a gold standard and guidance for researchers working on intelligent algorithms. The roots of universal induction have been laid exactly half-a-century ago and the roots of universal intelligence exactly one decade ago. So it is timely to take stock of what has been achieved and what remains to be done. Since there are already good recent surveys, I describe the state-of-the-art only in passing and refer the reader to the literature.

Partially observable Markov decision processes (POMDPs) have been successfully applied to various robot motion planning tasks under uncertainty. However, most existing POMDP algorithms assume a discrete state space, while the natural state space of a robot is often continuous. This paper presents Monte Carlo Value Iteration (MCVI) for continuous-state POMDPs. MCVI samples both a robot’s state space and the corresponding belief space, and avoids inefficient a priori discretization of the state space as a grid. Both theoretical results and preliminary experimental results indicate that MCVI is a promising new approach for robot motion planning under uncertainty.

Exploration for robotic mapping is typically handled using greedy entropy reduction. Here we show how to apply information lookahead planning to a challenging instance of this problem in which an Autonomous Underwater Vehicle (AUV) maps hydrothermal vents. Given a simulation of vent behaviour we derive an observation function to turn the planning for mapping problem into a POMDP. We test a variety of information state MDP algorithms against greedy, systematic and reactive search strategies. We show that directly rewarding the AUV for visiting vents induces effective mapping strategies. We evaluate the algorithms in simulation and show that our information lookahead method outperforms the others. 1