Showing papers by "Sebastian Thrun published in 2004"

Simultaneous Localization and Mapping with Sparse Extended Information Filters

Abstract: In this paper we describe a scalable algorithm for the simultaneous mapping and localization (SLAM) problem. SLAM is the problem of acquiring a map of a static environment with a mobile robot. The vast majority of SLAM algorithms are based on the extended Kalman filter (EKF). In this paper we advocate an algorithm that relies on the dual of the EKF, the extended information filter (EIF). We show that when represented in the information form, map posteriors are dominated by a small number of links that tie together nearby features in the map. This insight is developed into a sparse variant of the EIF, called the sparse extended information filter (SEIF). SEIFs represent maps by graphical networks of features that are locally interconnected, where links represent relative information between pairs of nearby features, as well as information about the robot’s pose relative to the map. We show that all essential update equations in SEIFs can be executed in constant time, irrespective of the size of the map. We...

6D SLAM with an application in autonomous mine mapping

Abstract: To create with an autonomous mobile robot a 3D volumetric map of a scene it is necessary to gage several 3D scans and to merge them into one consistent 3D model. This paper provides a new solution to the simultaneous localization and mapping (SLAM) problem with six degrees of freedom. Robot motion on natural surfaces has to cope with yaw, pitch and roll angles, turning pose estimation into a problem in six mathematical dimensions. A fast variant of the Iterative Closest Points algorithm registers the 3D scans in a common coordinate system and relocalizes the robot. Finally, consistent 3D maps are generated using a global relaxation. The algorithms have been tested with 3D scans taken in the Mathies mine, Pittsburgh, PA. Abandoned mines pose significant problems to society, yet a large fraction of them lack accurate 3D maps.

The Correlated Correspondence Algorithm for Unsupervised Registration of Nonrigid Surfaces

Abstract: We present an unsupervised algorithm for registering 3D surface scans of an object undergoing significant deformations. Our algorithm does not need markers, nor does it assume prior knowledge about object shape, the dynamics of its deformation, or scan alignment. The algorithm registers two meshes by optimizing a joint probabilistic model over all point-to-point correspondences between them. This model enforces preservation of local mesh geometry, as well as more global constraints that capture the preservation of geodesic distance between corresponding point pairs. The algorithm applies even when one of the meshes is an incomplete range scan; thus, it can be used to automatically fill in the remaining surfaces for this partial scan, even if those surfaces were previously only seen in a different configuration. We evaluate the algorithm on several real-world datasets, where we demonstrate good results in the presence of significant movement of articulated parts and non-rigid surface deformation. Finally, we show that the output of the algorithm can be used for compelling computer graphics tasks such as interpolation between two scans of a non-rigid object and automatic recovery of articulated object models.

Toward a framework for human-robot interaction

Abstract: The goal of this article is to introduce the reader to the rich and vibrant field of robotics. Robotics is a field in change; the meaning of the term robot today differs substantially from the term just 1 decade ago. The primary purpose of this article is to provide a comprehensive description of past- and present-day robotics. It identifies the major epochs of robotic technology and systems--from industrial to service robotics-and characterizes the different styles of human-robot interaction paradigmatic for each epoch. To help set the agenda for research on human-robot interaction, the article articulates some of the most pressing open questions pertaining to modern-day human-robot interaction.

Autonomous exploration and mapping of abandoned mines

Abstract: This article discusses the software architecture of an autonomous robotic system designed to explore and map abandoned mines. A new set of software tools is presented, enabling robots to acquire maps of unprecedented size and accuracy. On 30 May 2003, the robot "Groundhog" successfully explored and mapped a main corridor of the abandoned Mathies mine near Courtney, Pennsylvania. This article also discusses some of the challenges that arise in the subterranean environments and some the difficulties of building truly autonomous robots.

Real-time fault diagnosis [robot fault diagnosis]

Abstract: This article presents a number of complementary algorithms for detecting faults on-board operating robots, where a fault is defined as a deviation from expected behavior. The algorithms focus on faults that cannot directly be detected from current sensor values but require inference from a sequence of time-varying sensor values. Each algorithm provides an independent improvement over the basic approach. These improvements are not mutually exclusive, and the algorithms may be combined to suit the application domain. All the approaches presented require dynamic models representing the behavior of each of the fault and operational states. These models can be built from analytical models of the robot dynamics, data from simulation, or from the real robot. All the approaches presented detect faults from a finite number of known fault conditions, although there may potentially be a very large number of these faults.

Approximate Solutions for Partially Observable Stochastic Games with Common Payoffs

Abstract: Partially observable decentralized decision making in robot teams is fundamentally different from decision making in fully observable problems. Team members cannot simply apply single-agent solution techniques in parallel. Instead, we must turn to game theoretic frameworks to correctly model the problem. While partially observable stochastic games (POSGs) provide a solution model for decentralized robot teams, this model quickly becomes intractable. We propose an algorithm that approximates POSGs as a series of smaller, related Bayesian games, using heuristics such as QMDP to provide the future discounted value of actions. This algorithm trades off limited look-ahead in uncertainty for computational feasibility, and results in policies that are locally optimal with respect to the selected heuristic. Empirical results are provided for both a simple problem for which the full POSG can also be constructed, as well as more complex, robot-inspired, problems.

A complete algorithm for visibility-based pursuit-evasion with limited field of view

Abstract: We study a form of the pursuit-evasion problem, in which one or more searchers must move through a given environment so as to guarantee detection of any and all evaders, which can move arbitrarily fast. Our goal is to develop techniques for coordinating teams of robots to execute this task in application domains such as clearing a building, for reasons of security or safety. To this end, we introduce a new class of searcher, the -searcher, which can be readily instantiated as a physical mobile robot. We present a detailed analysis of the pursuit-evasion problem using -searchers. We show that computing the minimum number of -searchers required to search a given environment is NP-hard, and present the first complete search algorithm for a single -searcher. We show how this algorithm can be extended to handle multiple searchers, and give examples of computed trajectories.

FastSLAM: An Efficient Solution to the Simultaneous Localization And Mapping Problem with Unknown Data

Abstract: This article provides a comprehensive description of FastSLAM, a new family of algorithms for the simultaneous localization and mapping problem, which specifically address hard data association problems. The algorithm uses a particle filter for sampling robot paths, and extended Kalman filters for representing maps acquired by the vehicle. This article presents two variants of this algorithm, the original algorithm along with a more recent variant that provides improved performance in certain operating regimes. In addition to a mathematical derivation of the new algorithm, we present a proof of convergence and experimental results on its performance on real-world data.

Detecting and modeling doors with mobile robots

Abstract: We describe a probabilistic framework for detection and modeling of doors from sensor data acquired in corridor environments with mobile robots. The framework captures shape, color, and motion properties of door and wall objects. The probabilistic model is optimized with a version of the expectation maximization algorithm, which segments the environment into door and wall objects and learns their properties. The framework allows the robot to generalize the properties of detected object instances to new object instances. We demonstrate the algorithm on real-world data acquired by a Pioneer robot equipped with a laser range finder and an omni-directional camera. Our results show that our algorithm reliably segments the environment into walls and doors, finding both doors that move and doors that do not move. We show that our approach achieves better results than models that only capture behavior, or only capture appearance.

A real-time expectation-maximization algorithm for acquiring multiplanar maps of indoor environments with mobile robots.

Abstract: This paper presents a real-time algorithm for acquiring compact three-dimensional maps of indoor environments, using a mobile robot equipped with range and imaging sensors. Building on previous work on real-time pose estimation during mapping, our approach extends the popular expectation-maximization algorithm to multisurface models, and makes it amenable to real-time execution. Maps acquired by our algorithm consist of compact sets of textured polygons that can be visualized interactively. Experimental results obtained in corridor-type environments illustrate that compact and accurate maps can be acquired in real time and in a fully automated fashion.

A real-time expectation-maximization algorithm for acquiring multiplanar maps of indoor environments with mobile robots

Abstract: This paper presents a real-time algorithm for acquiring compact three-dimensional maps of indoor environments, using a mobile robot equipped with range and imaging sensors. Building on previous work on real-time pose estimation during mapping, our approach extends the popular expectation-maximization algorithm to multisurface models, and makes it amenable to real-time execution. Maps acquired by our algorithm consist of compact sets of textured polygons that can be visualized interactively. Experimental results obtained in corridor-type environments illustrate that compact and accurate maps can be acquired in real time and in a fully automated fashion.

Learning low dimensional predictive representations

Abstract: Predictive state representations (PSRs) have recently been proposed as an alternative to partially observable Markov decision processes (POMDPs) for representing the state of a dynamical system (Littman et al., 2001). We present a learning algorithm that learns a PSR from observational data. Our algorithm produces a variant of PSRs called transformed predictive state representations (TPSRs). We provide an efficient principal-components-based algorithm for learning a TPSR, and show that TPSRs can perform well in comparison to Hidden Markov Models learned with Baum-Welch in a real world robot tracking task for low dimensional representations and long prediction horizons.

Recovering articulated object models from 3D range data

Abstract: We address the problem of unsupervised learning of complex articulated object models from 3D range data. We describe an algorithm whose input is a set of meshes corresponding to different configurations of an articulated object. The algorithm automatically recovers a decomposition of the object into approximately rigid parts, the location of the parts in the different object instances, and the articulated object skeleton linking the parts. Our algorithm first registers all the meshes using an unsupervised non-rigid technique described in a companion paper. It then segments the meshes using a graphical model that captures the spatial contiguity of parts. The segmentation is done using the EM algorithm, iterating between finding a decomposition of the object into rigid parts, and finding the location of the parts in the object instances. Although the graphical model is densely connected, the object decomposition step can be performed optimally and efficiently, allowing us to identify a large number of object parts while avoiding local maxima. We demonstrate the algorithm on real world datasets, recovering a 15-part articulated model of a human puppet from just 7 different puppet configurations, as well as a 4 part model of a flexing arm where significant non-rigid deformation was present.

Simultaneous Mapping and Localization with Sparse Extended Information Filters: Theory and Initial Results

Abstract: This paper describes a scalable algorithm for the simultaneous mapping and localization (SLAM) problem. SLAM is the problem of determining the location of environmental features with a roving robot. Many of today’s popular techniques are based on extended Kalman filters (EKFs), which require update time quadratic in the number of features in the map. This paper develops the notion of sparse extended information filters (SEIFs), as a new method for solving the SLAM problem. SEIFs exploit structure inherent in the SLAM problem, representing maps through local, Web-like networks of features. By doing so, updates can be performed in constant time, irrespective of the number of features in the map. This paper presents several original constant-time results of SEIFs, and provides simulation results that show the high accuracy of the resulting maps in comparison to the computationally more cumbersome EKF solution.

A campaign in autonomous mine mapping

Abstract: Unknown, unexplored and abandoned subterranean voids threaten mining operations, surface developments and the environment. Hazards within these spaces preclude human access to create and verify extensive maps or to characterize and analyze the environment. To that end, we have developed a mobile robot capable of autonomously exploring and mapping abandoned mines. To operate without communications in a harsh environment with little chance of rescue, this robot must have a robust electro-mechanical platform, a reliable software system, and a dependable means of failure recovery. Presented are the mechanisms, algorithms, and analysis tools that enable autonomous mine exploration and mapping along with extensive experimental results from eight successful deployments into the abandoned Mathies coal mine near Pittsburgh, PA.

Real-time fault detection and situational awareness for rovers: report on the Mars technology program task

Abstract: In this paper we describe the results of a project funded by the Mars technology program at NASA, aimed at developing algorithms to meet this requirement. We describe a number of particle filtering-based algorithms for state estimation which we have demonstrated successfully on diagnosis problems including the K-9 rover at NASA Ames Research Center and the Hyperion rover at CMU. Due to the close interaction between a rover and its environment, traditional discrete approaches to diagnosis are impractical for this domain. Therefore we model rover subsystems as hybrid discrete/continuous systems. There are three major challenges to make particle filters work in this domain. The first is that fault states typically have a very low probability of occurring, so there is a risk that no samples enter fault states. The second issue is coping with the high-dimensional continuous state spaces of the hybrid system models, and the third is the severely constrained computational power available on the rover. This means that very few samples can be used if we wish to track the system state in real time. We describe a number of approaches to rover diagnosis specifically designed to address these challenges.

A multi-resolution pyramid for outdoor robot terrain perception

Abstract: This paper addresses the problem of outdoor terrain modeling for the purposes of mobile robot navigation. We propose an approach in which a robot acquires a set of terrain models at differing resolutions. Our approach addresses one of the major shortcomings of Bayesian reasoning when applied to terrain modeling, namely artifacts that arise from the limited spatial resolution of robot perception. Limited spatial resolution causes small obstacles to be detectable only at close range. Hence, a Bayes filter estimating the state of terrain segments must consider the ranges at which that terrain is observed. We develop a multi-resolution approach that maintains multiple navigation maps, and derive rational arguments for the number of layers and their resolutions. We show that our approach yields significantly better results in a practical robot system, capable of acquiring detailed 3-D maps in large-scale outdoor environments.

Tractable planning under uncertainty: exploiting structure

Abstract: The problem of planning under uncertainty has received significant attention in the scientific community over the past few years. It is now well-recognized that considering uncertainty during planning and decision-making is imperative to the design of robust computer systems. This is particularly crucial in robotics, where the ability to interact effectively with real-world environments is a prerequisite for success. The Partially Observable Markov Decision Process (POMDP) provides a rich framework for planning under uncertainty. The POMDP model can optimize sequences of actions which are robust to sensor noise, missing information, occlusion, as well as imprecise actuators. While the model is sufficiently rich to address most robotic planning problems, exact solutions are generally intractable for all but the smallest problems. This thesis argues that large POMDP problems can be solved by exploiting natural structural constraints. In support of this, we propose two distinct but complementary algorithms which overcome tractability issues in POMDP planning. PBVI is a sample-based approach which approximates a value function solution by planning over a small number of salient information states. PolCA+ is a hierarchical approach which leverages structural properties of a problem to decompose it into a set of smaller, easy-to-solve, problems. These techniques improve the tractability of POMDP planning to the point where POMDP-based robot controllers are a reality. This is demonstrated through the successful deployment of a nursing assistant robot.

A passive approach to sensor network localization

Abstract: Sensor networks present the opportunity to accurately localize the phenomena of interest. To be able to do so however, sensor nodes, need themselves to be accurately localized. We present an algorithm to do this based on uncontrolled environmental sounds observed by each of the sensor nodes. A probabilistic generative model is presented and it is shown that the sensor node localization problem is equivalent to maximum likelihood estimation in the model. Experimental results are presented for both simulated sensor nodes and Crossbow MICA2 sensor nodes.

Particle Filters for Rover Fault Diagnosis

Abstract: states (see Figure 2). In addition to the physical state set D, the variable resolution model consists of a set of M abstract states { } ) ( ) 1 ( M a a A = that represent sets of states and/or other abstract states: = ) i ( i ) k ( j a d a A (somewhat simplified) algorithm for abstracting and refining states in the VRPF is shown in Figure 3. Figure 3: Variable resolution particle filter algorithm. [Equation 8] The initial set of particles { } N 1 i ] i 0 ] i 0 0 x , a B = = are drawn from the prior distribution. R0 is set to the set of unique states (physical or abstract) represented in B0. The particle set, Bt is then recursively drawn from Bt-1 as follows: 1. Project all the particles to physical states to use the physical transition and measurement models. If a particle, 1 t ) j ( ] i [ 1 t B d a − − ∈ = , is in an abstract state, one of its descendant physical states } d { ) j ( , is selected proportional to the prior probability of the physical states as follows:

PAO for planning with hidden state

Abstract: We describe a heuristic search algorithm for generating optimal plans in a new class of decision problem, characterised by the incorporation of hidden state. The approach exploits the nature of the hidden state to reduce the state space by orders of magnitude. It then interleaves heuristic expansion of the reduced space with forwards and backwards propagation phases to produce a solution in a fraction of the time required by other techniques. Results are provided on an outdoor path planning application.

Simultaneous localization and mapping with active stereo vision

Abstract: We present an algorithm for creating globally consistent three-dimensional maps from depth fields produced by camera-based range measurement systems. Our approach is specifically suited to dealing with the high noise levels and the large number of outliers often produced by such systems. Range data is filtered to reject outliers within each scan. The point-to-plane variant of ICP is used for local alignment, including weightings that favor nearby points and a novel outlier rejection strategy that increases the robustness for this class of data while eliminating the burden of user-specified thresholds. Global consistency is imposed on cycles by optimally distributing the cyclic discrepancy according to the local fit correlation matrices. The algorithm is demonstrated on a dataset collected by an active unstructured-light space-time stereo vision system.

A probabilistic approach to inference with limited information in sensor networks

Abstract: We present a methodology for a sensor network to answer queries with limited and stochastic information using probabilistic techniques. This capability is useful in that it allows sensor networks to answer queries effectively even when present information is partially corrupt and when more information is unavailable or too costly to obtain. We use a Bayesian network to model the sensor network and Markov chain Monte Carlo sampling to perform approximate inference. We demonstrate our technique on the specific problem of determining whether a friendly agent is surrounded by enemy agents and present simulation results for it.

Learning user models of mobility-related activities through instrumented walking aids

Abstract: We present a robotic walking aid capable of learning models of users' walking-related activities. Our walker is instrumented to provide guidance to elderly people when navigating their environments; however, such guidance is difficult to provide without knowing what activity a person is engaged in (e.g., where a person wants to go). The main contribution of this paper is an algorithm for learning models of users of the walker. These models are defined at multiple levels of abstractions, and learned from actual usage data using statistical techniques. We demonstrate that our approach succeeds in determining the specific activity in which a user engages when using the walker. One of our proto-type walkers was tested in an assisted living facility near Pittsburgh, PA; a more recent model was extensively evaluated in a university environment.

Planning for Markov Decision Processes with Sparse Stochasticity

Abstract: Planning algorithms designed for deterministic worlds, such as A* search, usually run much faster than algorithms designed for worlds with uncertain action outcomes, such as value iteration. Real-world planning problems often exhibit uncertainty, which forces us to use the slower algorithms to solve them. Many real-world planning problems exhibit sparse uncertainty: there are long sequences of deterministic actions which accomplish tasks like moving sensor platforms into place, interspersed with a small number of sensing actions which have uncertain outcomes. In this paper we describe a new planning algorithm, called MCP (short for MDP Compression Planning), which combines A* search with value iteration for solving Stochastic Shortest Path problem in MDPs with sparse stochasticity. We present experiments which show that MCP can run substantially faster than competing planners in domains with sparse uncertainty; these experiments are based on a simulation of a ground robot cooperating with a helicopter to fill in a partial map and move to a goal location.

Visibility-based pursuit-evasion with limited field of view

Abstract: We study a form of the pursuit-evasion problem, in which one or more searchers must move through a given environment so as to guarantee detection of any and all evaders, which can move arbitrarily fast. Our goal is to develop techniques for coordinating teams of robots to execute this task in application domains such as clearing a building, for reasons of security or safety. To this end, we introduce a new class of searcher, the Φ-searcher, which can be readily instantiated as a physical mobile robot. We present a detailed analysis of the pursuit-evasion problem using Φ-searchers. We show that computing the minimum number of Φ-searchers required to search a given environment is NP-hard, and present the first complete search algorithm for a single Φ-searcher. We show how this algorithm can be extended to handle multiple searchers, and give examples of computed trajectories.

Tractable Particle Filters for Robot Fault Diagnosis

Abstract: EXPERIENCE has shown that even carefully designed and tested robots may encounter anomalous situations. It is therefore important for robots to monitor their state so that anomalous situations may be detected in a timely manner. Robot fault diagnosis typically requires tracking a very large number of possible faults in complex non-linear dynamic systems with noisy sensors. Traditional methods either ignore the uncertainly or use linear approximations of nonlinear system dynamics. Such approximations are often unrealistic, and as a result faults either go undetected or become confused with non-fault conditions. Probability theory provides a natural representation for uncertainty, but an exact Bayesian solution for the diagnosis problem is intractable. Monte Carlo approximations have demonstrated considerable success in application domains such as computer vision and robot localization and mapping. But, classical Monte Carlo methods, such as particle filters, can suffer from substantial computational complexity. This is particularly true with the presence of rare, yet important events, such as many system faults. The thesis makes contributions to the theory of probabilistic state-estimation to solve this problem. It presents a novel approach that outperforms existing algorithms (in terms of computational efficiency) for reliably estimating the state of general (nonlinear, nonGaussian) dynamic systems in real-time in the presence of uncertainty (including rare events such as faults). The thesis presents a set of complementary algorithms that provide an approach for computationally tractable fault diagnosis. These algorithms leverage probabilistic approaches to decision theory and information theory to efficiently track a large number of faults in a general dynamic system with noisy measurements. The problem of fault diagnosis is represented as hybrid (discrete/continuous) state estimation. Taking advantage of structure in the domain it dynamically concentrates computation in the regions of state space that are currently most relevant without losing track of less likely states. Experiments with a dynamic simulation of a six-wheel rocker-bogie rover show a significant improvement in performance over the classical approach.