Showing papers by "Sebastian Thrun published in 2000"

Text Classification from Labeled and Unlabeled Documents using EM

Abstract: This paper shows that the accuracy of learned text classifiers can be improved by augmenting a small number of labeled training documents with a large pool of unlabeled documents. This is important because in many text classification problems obtaining training labels is expensive, while large quantities of unlabeled documents are readily available. We introduce an algorithm for learning from labeled and unlabeled documents based on the combination of Expectation-Maximization (EM) and a naive Bayes classifier. The algorithm first trains a classifier using the available labeled documents, and probabilistically labels the unlabeled documents. It then trains a new classifier using the labels for all the documents, and iterates to convergence. This basic EM procedure works well when the data conform to the generative assumptions of the model. However these assumptions are often violated in practice, and poor performance can result. We present two extensions to the algorithm that improve classification accuracy under these conditions: (1) a weighting factor to modulate the contribution of the unlabeled data, and (2) the use of multiple mixture components per class. Experimental results, obtained using text from three different real-world tasks, show that the use of unlabeled data reduces classification error by up to 30%.

Collaborative multi-robot exploration

Abstract: In this paper we consider the problem of exploring an unknown environment by a team of robots. As in single-robot exploration the goal is to minimize the overall exploration time. The key problem to be solved therefore is to choose appropriate target points for the individual robots so that they simultaneously explore different regions of their environment. We present a probabilistic approach for the coordination of multiple robots which, in contrast to previous approaches, simultaneously takes into account the costs of reaching a target point and the utility of target points. The utility of target points is given by the size of the unexplored area that a robot can cover with its sensors upon reaching a target position. Whenever a target point is assigned to a specific robot, the utility of the unexplored area visible from this target position is reduced for the other robots. This way, a team of multiple robots assigns different target points to the individual robots. The technique has been implemented and tested extensively in real-world experiments and simulation runs. The results given in this paper demonstrate that our coordination technique significantly reduces the exploration time compared to previous approaches.

A real-time algorithm for mobile robot mapping with applications to multi-robot and 3D mapping

Abstract: We present an incremental method for concurrent mapping and localization for mobile robots equipped with 2D laser range finders. The approach uses a fast implementation of scan-matching for mapping, paired with a sample-based probabilistic method for localization. Compact 3D maps are generated using a multi-resolution approach adopted from the computer graphics literature, fed by data from a dual laser system. Our approach builds 3D maps of large, cyclic environments in real-time, and it is robust. Experimental results illustrate that accurate maps of large, cyclic environments can be generated even in the absence of any odometric data.

A Probabilistic Approach to Collaborative Multi-Robot Localization

Abstract: This paper presents a statistical algorithm for collaborative mobile robot localization. Our approach uses a sample-based version of Markov localization, capable of localizing mobile robots in an any-time fashion. When teams of robots localize themselves in the same environment, probabilistic methods are employed to synchronize each robot's belief whenever one robot detects another. As a result, the robots localize themselves faster, maintain higher accuracy, and high-cost sensors are amortized across multiple robot platforms. The technique has been implemented and tested using two mobile robots equipped with cameras and laser range-finders for detecting other robots. The results, obtained with the real robots and in series of simulation runs, illustrate drastic improvements in localization speed and accuracy when compared to conventional single-robot localization. A further experiment demonstrates that under certain conditions, successful localization is only possible if teams of heterogeneous robots collaborate during localization.

Coordination for Multi-Robot Exploration and Mapping

Abstract: This paper addresses the problem of exploration and mapping of an unknown environment by multiple robots. The mapping algorithm is an on-line approach to likelihood maximization that uses hill climbing to find maps that are maximally consistent with sensor data and odometry. The exploration algorithm explicitly coordinates the robots. It tries to maximize overall utility by minimizing the potential for overlap in information gain amongst the various robots. For both the exploration and mapping algorithms, most of the computations are distributed. The techniques have been tested extensively in real-world trials and simulations. The results demonstrate the performance improvements and robustness that accrue from our multirobot approach to exploration.

Probabilistic Algorithms and the Interactive Museum Tour-Guide Robot Minerva

Abstract: This paper describes Minerva, an interactive tour-guide robot that was successfully deployed in a Smithsonian museum. Minerva’s software is pervasively probabilistic, relying on explicit representations of uncertainty in perception and control. This article describes Minerva’s major software components, and provides a comparative analysis of the results obtained in the Smithsonian museum. During two weeks of highly successful operation, the robot interacted with thousands of people, both in the museum and through the Web, traversing more than 44km at speeds of up to 163 cm/sec in the unmodie d museum.

Probabilistic Algorithms in Robotics

Abstract: This article describes a methodology for programming robots known as probabilistic robotics The probabilistic paradigm pays tribute to the inherent uncertainty in robot perception, relying on explicit representations of uncertainty when determining what to do This article surveys some of the progress in the field, using in-depth examples to illustrate some of the nuts and bolts of the basic approach My central conjecture is that the probabilistic approach to robotics scales better to complex real-world applications than approaches that ignore a robot's uncertainty

A Gesture Based Interface for Human-Robot Interaction

Abstract: Service robotics is currently a highly active research area in robotics, with enormous societal potential. Since service robots directly interact with people, finding “natural” and easy-to-use user interfaces is of fundamental importance. While past work has predominately focussed on issues such as navigation and manipulation, relatively few robotic systems are equipped with flexible user interfaces that permit controlling the robot by “natural” means. This paper describes a gesture interface for the control of a mobile robot equipped with a manipulator. The interface uses a camera to track a person and recognize gestures involving arm motion. A fast, adaptive tracking algorithm enables the robot to track and follow a person reliably through office environments with changing lighting conditions. Two alternative methods for gesture recognition are compared: a template based approach and a neural network approach. Both are combined with the Viterbi algorithm for the recognition of gestures defined through arm motion (in addition to static arm poses). Results are reported in the context of an interactive clean-up task, where a person guides the robot to specific locations that need to be cleaned and instructs the robot to pick up trash.

Structure from motion without correspondence

Abstract: A method is presented to recover 3D scene structure and camera motion from multiple images without the need for correspondence information. The problem is framed as finding the maximum likelihood structure and motion given only the 2D measurements, integrating over all possible assignments of 3D features to 2D measurements. This goal is achieved by means of an algorithm which iteratively refines a probability distribution over the set of all correspondence assignments. At each iteration a new structure from motion problem is solved, using as input a set of 'virtual measurements' derived from this probability distribution. The distribution needed can be efficiently obtained by Markov Chain Monte Carlo sampling. The approach is cast within the framework of Expectation-Maximization, which guarantees convergence to a local maximizer of the likelihood. The algorithm works well in practice, as will be demonstrated using results on several real image sequences.

Spoken dialogue management using probabilistic reasoning

Abstract: Spoken dialogue managers have benefited from using stochastic planners such as Markov Decision Processes (MDPs). However, so far, MDPs do not handle well noisy and ambiguous speech utterances. We use a Partially Observable Markov Decision Process (POMDP)-style approach to generate dialogue strategies by inverting the notion of dialogue state; the state represents the user's intentions, rather than the system state. We demonstrate that under the same noisy conditions, a POMDP dialogue manager makes fewer mistakes than an MDP dialogue manager. Furthermore, as the quality of speech recognition degrades, the POMDP dialogue manager automatically adjusts the policy.

Decision-Theoretic, High-Level Agent Programming in the Situation Calculus

Abstract: We propose a framework for robot programming which allows the seamless integration of explicit agent programming with decision-theoretic planning. Specifically, the DTGologmodel allows one to partially specify a control program in a highlevel, logical language, and provides an interpreter that, given a logical axiomatization of a domain, will determine the optimal completion of that program (viewed as a Markov decision process). We demonstrate the utility of this model with results obtained in an officedelivery robotics domain.

Monte Carlo Localization with Mixture Proposal Distribution

Abstract: Monte Carlo localization (MCL) is a Bayesian algorithm for mobile robot localization based on particle filters, which has enjoyed great practical success. This paper points out a limitation of MCL which is counter-intuitive, namely that better sensors can yield worse results. An analysis of this problem leads to the formulation of a new proposal distribution for the Monte Carlo sampling step. Extensive experimental results with physical robots suggest that the new algorithm is significantly more robust and accurate than plain MCL. Obviously, these results transcend beyond mobile robot localization and apply to a range of particle filter applications.

Web interfaces for mobile robots in public places

Abstract: The growth of the World Wide Web provides unique opportunities to bring robots closer to people. The vision behind such endeavors ranges from relatively simple web-based inspections and surveillance applications to highly versatile applications that use robots connected to the web to establish a remote telepresence in dynamic and populated environments. In the latter scenario, robots play the role of a physical mediator, enabling remote people to acquire information, explore, manipulate, communicate, and interact physically with people far away. The article describes a series of web interfaces designed to remotely operate mobile robots in public places through the web. The design of these interfaces specifically addresses issues such as low bandwidth of interconnections, control brokering, and shared control as well as interaction with people in the robot's environment, which arise naturally in applications with web-based robot control. The interfaces have been tested extensively using two deployed service robots, which were installed as interactive tour guides in two museums. The article also discusses trade-offs and limitations of web-based robots that interact with people in populated public places.

Coordinated deployment of multiple, heterogeneous robots

Abstract: To be truly useful, mobile robots need to be fairly autonomous and easy to control. This is especially true in situations where multiple robots are used, due to the increase in sensory information and the fact that the robots can interfere with one another. The paper describes a system that integrates autonomous navigation, a task executive, task planning, and an intuitive graphical user interface to control multiple, heterogeneous robots. We have demonstrated a prototype system that plans and coordinates the deployment of teams of robots. Testing has shown the effectiveness and robustness of the system, and of the coordination strategies in particular.

An Online Mapping Algorithm for Teams of Mobile Robots

Abstract: : We propose a new probabilistic algorithm for online mapping of unknown environments with teams of robots. At the core of the algorithm is a technique that combines fast maximum likelihood map growing with a Monte Carlo localizer that uses particle representations. The combination of both yields an online algorithm that can cope with large odometric errors typically found when mapping an environment with cycles. The algorithm can be implemented distributedly on multiple robot platforms, enabling a team of robots to cooperatively generate a single map of their environment. Finally, an extension is described for acquiring three-dimensional maps, which capture the structure and visual appearance of indoor environments in 3D.

Towards programming tools for robots that integrate probabilistic computation and learning

Abstract: This paper describes a programming language extension of C++, called CES, specifically targeted towards mobile robot control. CES's design is motivated by a recent series of successful probabilistic methods for mobile robot control, with the goal of facilitating the development of such probabilistic software in future robot applications. CES extends C++ by two ideas: Computing with probability distributions, and built-in mechanisms for learning from examples as a new means of programming. An example program, used to control a mail-delivering robot with gesture command interface, illustrates that CES may reduce the code development by two orders of magnitude. CES differs from other special-purpose programming languages in the field, which typically emphasize concurrency and real-time/event-driven processing.

Feature Correspondence: A Markov Chain Monte Carlo Approach

Abstract: When trying to recover 3D structure from a set of images, the most difficult problem is establishing the correspondence between the measurements. Most existing approaches assume that features can be tracked across frames, whereas methods that exploit rigidity constraints to facilitate matching do so only under restricted camera motion. In this paper we propose a Bayesian approach that avoids the brittleness associated with singling out one "best" correspondence, and instead consider the distribution over all possible correspondences. We treat both a fully Bayesian approach that yields a posterior distribution, and a MAP approach that makes use of EM to maximize this posterior. We show how Markov chain Monte Carlo methods can be used to implement these techniques in practice, and present experimental results on real data.

A Review of Reinforcement Learning

Abstract: Review of "Reinforcement Learning: An Introduction, Richard S. Sutton and Andrew G. Barto, The MIT Press, Cambridge, Massachusetts, 1998, 322 pp., ISBN 0-262-19398-1.

Efficient Multi-Robot Localization Based on Monte Carlo Approximation

Abstract: This paper presents a probabilistic algorithm for collaborative mobile robot localization. Our approach uses a sample-based version of Markov localization, capable of localizing mobile robots in any-time fashion. When teams of robots localize themselves in the same environment, probabilistic methods are employed to synchronize each robot’s belief whenever one robot detects another. As a result, the robots localize themselves faster, maintain higher accuracy, and high-cost sensors are amortized across multiple robot platforms. The paper also describes experimental results obtained using two mobile robots, using computer vision and laser range-finding for detecting each other and estimating each other’s relative location. The results, obtained in an indoor office environment, illustrate drastic improvements in localization speed and accuracy when compared to conventional single-robot localization.