| Infant-Like Social Interactions Between a Robot and a Human Caregiver |
2006 |
58 pages |
| Authors:
Cynthia Breazael; Brian Scassellati; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
 | From birth human infants are immersed in a social environment that allows them to learn by leveraging the skills and capabilities of their caregivers. A critical pre-cursor to this type of social learning is the ability to maintain interaction levels that are neither overwhelming nor under-stimulating. In this paper we present a mechanism for an autonomous robot to regulate the intensity of its social interactions with a human. Similar to ... |
|
| Discriminating Animate from Inanimate Visual Stimuli |
2006 |
7 pages |
| Authors:
Brian Scassellati; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| From as early as 6 months of age human children distinguish between motion patterns generated by animate objects from patterns generated by moving inanimate objects even when the only stimulus that the child observes is a single point of light moving against a blank background. The mechanisms by which the animate/inanimate distinction are made are unknown, but have been shown to rely only upon the spatial and temporal properties of ... |
|
| Orientation Behavior Using Registered Topographic Maps |
2006 |
11 pages |
| Authors:
Cynthia Ferrell; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The ability to orient toward visual auditory or tactile stimuli is an important skill for systems intended to interact with and explore their environment. In the brain of mammalian vertebrates the Superior Colliculus is specialized for integrating multi-modal sensory information and for using this information to orient the animal to the source sensory stimuli such as noisy, moving objects. Within the Superior Colliculus this ability appears to be implemented using ... |
|
| Extracting Answers from the Web Using Knowledge Annotation and Knowledge Mining Techniques |
2006 |
11 pages |
| Authors:
Jimmy Lin; Aaron Fernandes; Boris Katz; Gregory Marton; Stefanie Tellex; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Aranea is a question answering system that extracts answers from the World Wide Web using knowledge annotation and knowledge mining techniques. Knowledge annotation, which utilizes semistructured database techniques, is effective for answering large classes of commonly occurring questions. Knowledge mining, which utilizes statistical techniques, can leverage the massive amounts of data available on the Web to overcome many natural language processing challenges. Aranea integrates these two different paradigms of question ... |
|
| Balancing Multiple Sources of Reward in Reinforcement Learning |
2006 |
8 pages |
| Authors:
Christian R. Shelton; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| For many problems which would be natural for reinforcement learning, the reward signal is not a single scalar value but has multiple scalar components. Examples of such problems include agents with multiple goals and agents with multiple users. Creating a single reward value by combining the multiple components can throw away vital information and can lead to incorrect solutions. We describe the multiple reward source problem and discuss the problems ... |
|
| Learning Humanoid Arm Gestures |
2005 |
4 pages |
| Authors:
Bryan Adams; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| While biological motion control systems are generally simple and robust, their robotic analogs tend to be just the opposite. While functions has driven many of the colorful architectures to date, we feel that a biologically- inspired system for monitoring the energy consumption of virtual muscles can lead to the development of more humanoid motion and gesture. |
|
| Affective Interaction Between Humans and Robots |
2005 |
10 pages |
| Authors:
Cynthia Breazeal; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This paper explores the role of emotive responses in communicative behavior between robots and humans. Done properly, affective communication should be natural and intuitive for people to understand. This implies that the robot's emotive behavior should be life-like. The ability to establish and maintain a rich affective dynamic with people has placed important constraints on our robotic implementation. We present our framework, discuss how these constraints have been addressed, and ... |
|
| Design of a Compliant and Force Sensing Hand for a Humanoid Robot |
2005 |
5 pages |
| Authors:
Aaron Edsinger-Gonzales; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Robot manipulation tasks in unknown and unstructured environments can often be better addressed with hands that are capable of force-sensing and passive compliance. We describe the design of a compact four degree-of-freedom (DOF) hand that exhibits these properties. This hand is being developed for a new humanoid robot platform. Our hand contains four modular Force Sensing Compliant (FSC) actuators acting on three fingers. One actuator controls the spread between two ... |
|
| Theory of Mind...for a Robot |
2005 |
5 pages |
| Authors:
Brian Scassellati; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| One of the fundamental social skills for humans is a theory of other minds. This set of skills allows us to attribute beliefs, goals and desires to other individuals. To take part in normal human social dynamics, a robot must not only know about the properties of objects, but also the properties of animate agents in the world. This paper presents the theories of Leslie (1994) and Baron-Cohen (1995) on ... |
|
| A Simple and Scalable Force Actuator |
2005 |
5 pages |
| Authors:
Eduardo Torres-Jara; Jessica Banks; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| One problem posed by robots that must perform in unstructured environments and interact with humans is that of achieving high fidelity force control as well as compliance in a compact form factor. Force control, as opposed to only position control, allows for safe, graceful contact behavior. Having spring elements placed in series between an actuator and its output provides this desired compliance. It also provides low mechanical impedance, precise force ... |
|
| Designing a Humanoid Robot Face to Fulfill a Social Contract |
2005 |
1 pages |
| Authors:
Aaron Edsinger; Una-May O'Reilly; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| In building humanoid robots, we are also building in implicit expectations about the social abilities and interactions that the machine should exhibit. The morphology and aesthetic of the robot play a critical role in defining these expectations. Specifically, the robot face plays a paramount role in establishing the social competence of the machine, and in defining a social contract between the observer and the robot. We will discuss the development ... |
|
| Versatile Tiled-Processor Architectures: The Raw Approach |
30 SEP 2004 |
33 pages |
| Authors:
Rodric M. Rabbah; Ian Bratt; Anant Agarwal; Krste Asanovic; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This presentation will describe the Raw architecture, its implementation, and performance. We will focus on Raw's ability to support a diverse set of applications (ranging from desktop to embedded workloads) and multiple forms of parallelism (including instruction-level parallelism (ILP) for desktop applications, and stream parallelism for embedded computing) as represented by the VersaBench suite. We will also report detailed performance measurements that quantify the versatility of Raw compared to some ... |
|
| Language-Level Transactions for Modular Reliable Systems |
30 SEP 2004 |
37 pages |
| Authors:
C. S. Ananian; Martin Rinard; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The authors show how to integrate non-blocking transactions into an object-oriented language, "transactifying" existing code to fix existing concurrency bugs and using transactions for modular fault-tolerance, backtracking, exception-handling, and concurrency control in new programs. We propose the use of compiler-supported atomic blocks to specify synchronization. This is less error-prone than manual maintenance of a locking discipline: deadlocks may be introduced when locks are not acquired and released in a highly ... |
|
| Reusable PVS Proof Strategies for Proving Abstraction Properties of I/O Automata |
04 JUL 2004 |
15 pages |
| Authors:
Sayan Mitra; Myla Archer; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Recent modifications to PVS Support a new technique for defining abstraction properties relating automata in a clean and uniform way. This definition technique employs specification templates that can support development of generic high level PVS strategies that set up the standard subgoals of these abstraction proofs and then execute the standard initial proof steps for these subgoals. In this paper, we describe an abstraction specification technique and associated abstraction proof ... |
|
| Programming Technology for Molecular-Scale Computing |
MAY 2004 |
9 pages |
| Authors:
Hal Abelson; Gerald J. Sussman; Thomas F. Knight; Radhika Nagpal; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Progress in molecular electronics is beginning to yield the technology for creating structures that incorporate myriads of nanoscale computationally active units. These could be fabricated at almost no cost, provided (1) the individual units need not all work correctly; and (2) there is no need to manufacture precise geometrical arrangements of the units or precise interconnections among them. Programming such structures to perform useful computations is a significant challenge. The ... |
|
| Statistical Learning: Stability is Sufficient for Generalization and Necessary and Sufficient for Consistency of Empirical Risk Minimization |
JAN 2004 |
56 pages |
| Authors:
Sayan Mukherjee; Partha Niyogi; Tomaso Poggio; Ryan Rifkin; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Solutions of learning problems by Empirical Risk Minimization (ERM) -- and almost-ERM when the minimizer does not exist -- need to be consistent, so that they may be predictive. They also need to be well-posed in the sense of being stable, so that they might be used robustly. We propose a statistical form of leave-one-out stability, called CVEEE(loo) stability. Our main new results are two. We prove that for bounded ... |
|
| Map Building from Human-Computer Interactions |
2004 |
6 pages |
| Authors:
MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| On-line help from a human actor will be exploited to facilitate computer perception. This paper proposes and innovative real-time algorithm - running on an active vision head - to build 3D scene descriptions from human cues. The theory is supported by experimental results both for figure/ground segregation of typical heavy objects in a scene (such as furniture), and for 3D object/scene reconstruction. |
|
| Learning Task Sequences from Scratch: Applications to the Control of Tools and Toys by a Humanoid Robot |
2004 |
6 pages |
| Authors:
Artur M. Arsenio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The goal of this work is to build perceptual and motor control systems for a humanoid robot, starting from an infant's early ability for detecting repetitive or abruptly varying world events from human-robot interactions, and walking developmentally towards robust perception and learning. This paper presents strategies for learning task sequences from human- robot interaction cues. Demonstration by human teachers facilitates robot learning to recognize new objects, such as tools or ... |
|
| Dissociated Dipoles: Image Representation via Non-local Comparisons |
AUG 2003 |
16 pages |
| Authors:
Benjamin J. Balas; Pawan Sinha; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| A fundamental question in visual neuroscience is how to represent image structure. The most common representational schemes rely on differential operators that compare adjacent image regions. While well-suited to encoding local relationships, such operators have significant drawbacks. Specifically, each filter's span is confounded with the size of its sub-fields, making it difficult to compare small regions across large distances. We find that such long-distance comparisons are more tolerant to common ... |
|
| Direction Estimation of Pedestrian from Images |
AUG 2003 |
13 pages |
| Authors:
Hiroaki Simizu; Tomaso Poggio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The capability of estimating the walking direction of people would be useful in many applications such as those involving autonomous cars and robots. We introduce an approach for estimating the walking direction of people from images, based on learning the correct classification of a still image by using SVMs. We find that the performance of the system can be improved by classifying each image of a walking sequence and combining ... |
|
| A Biological Model of Object Recognition with Feature Learning |
JUN 2003 |
61 pages |
| Authors:
Jennifer Louie; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Previous biological models of object recognition in cortex have been evaluated using idealized scenes and have hard-coded features, such as the HMAX model by Riesenhuber and Poggio [10]. Because HMAX uses the same set of features for all object classes, it does not perform well in the task of detecting a target object in clutter. This thesis presents a new model that integrates learning of object-specific features with the HMAX. ... |
|
| Knowledge Based Collaboration Webs |
MAR 2003 |
50 pages |
| Authors:
Howard Shrobe; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The Knowledge Based Collaboration Web (KBCW) project at the MIT artificial Intelligence Laboratory (AI) exploited representations and techniques used in AI for research and development of a platform and tools to support collaboration. A KBCW consists of nodes and links, where nodes represent a document, or a fragment of a document with a coherent prepositional content and links represent relationships between the nodes. Both the nodes and links are made ... |
|
| Relative Contributions of Internal and External Features to Face Recognition |
MAR 2003 |
13 pages |
| Authors:
Izzat N. Jarudi; Pawan Sinha; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The central challenge in face recognition lies in understanding the role different facial features play in our judgments of identity. Notable in this regard are the relative contributions of the internal (eyes, nose and mouth) and external (hair and jaw-line) features. Past studies that have investigated this issue have typically used high-resolution images or good-quality line drawings as facial stimuli. The results obtained are therefore most relevant for understanding the ... |
|
| Perceptual Evaluation of Video-Realistic Speech |
FEB 2003 |
19 pages |
| Authors:
Gadi Geiger; Tony Ezzat; Tomaso Poggio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| With many visual speech animation techniques now available, there is a clear need for systematic perceptual evaluation schemes. We describe here our scheme and its application to a new video-realistic (potentially indistinguishable from real recorded video) visual-speech animation system, called Mary 101. Two types of experiments were performed: a) distinguishing visually between real and synthetic image-sequences of the same utterances, (Turing tests) and b) gauging visual speech recognition by comparing ... |
|
| Towards Pervasive Robotics |
2003 |
2 pages |
| Authors:
Artur M. Arsenio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| In the near future, pervasive robotics will require small, light, and cheap robots that exhibit complex behaviors. These demands led to the development of the M2-M4 Macaco project -- a robotic active vision head. Macaco is a portable system, capable of emulating the head of different creatures both aesthetically and functionally. It integrates mechanisms for social interactions, autonomous navigation, and object analysis. |
|
| Open Object Recognition for Humanoid Robots |
2003 |
2 pages |
| Authors:
Paul Fitzpatrick; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Robots must be able to adapt gracefully to frequent and dramatic changes in their workspace if they are to operate successfully in human-centered environments, as opposed to controlled industrial settings. At the MIT Humanoid Robotics Group, investigators are developing methods that permit their robots to deduce the structure of novel activities, adopt the vocabulary appropriate for communication about the task at hand, and learn about the appearance and behavior of ... |
|
| Object Segmentation through Human-Robot Interactions in the Frequency Domain |
2003 |
7 pages |
| Authors:
Artur M. Arsenio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This paper presents a new embodied approach for object segmentation by a humanoid robot. It relies on interactions with a human teacher that drives the robot through the process of segmenting objects from arbitrarily complex, nonstatic images. Objects from a large spectrum of different scenarios are successfully segmented by the proposed algorithms. The paper discusses embodied object segmentation; detection of events in the frequency domain, including event detection, tracking, and ... |
|
| Shoes as a Platform for Vision |
2003 |
4 pages |
| Authors:
Paul Fitzpatrick; Charles C. Kemp; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The authors explore the use of a shoe-mounted camera as a sensory system for wearable computing. They demonstrate tools useful for gait analysis, obstacle detection, and context recognition. Using only visual information, they detect periods of stability and motion during walking. In the stable phase, the foot can be assumed to be parallel to the ground plane. In this condition, the floor dominates the lower part of the camera's view, ... |
|
| Perception and Perspective in Robotics |
2003 |
6 pages |
| Authors:
Paul Fitzpatrick; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| To a robot, the world is a sea of ambiguity, in which it will sink or swim depending on the robustness of its perceptual abilities. But robust machine perception has proven difficult to achieve. This paper argues that robots must be given not just particular perceptual competencies, but the tools to forge those competencies out of raw physical experiences. Three important tools for extending a robot's perceptual abilities whose importance ... |
|
| The Whole World in Your Hand: Active and Interactive Segmentation |
2003 |
8 pages |
| Authors:
Artur Arsenio; Paul Fitzpatrick; Charles C. Kemp; Giorgio Metta; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Object segmentation is a fundamental problem in computer vision and a powerful resource for development. This paper presents three embodied approaches to the visual segmentation of objects. Each approach to segmentation is aided by the presence of a hand or arm in the proximity of the object to be segmented. The first approach is suitable for a robotic system, where the robot can use its arm to evoke object motion. ... |
|
| Idempotent Vector Design for Standard Assembly of Biobricks |
2003 |
12 pages |
| Authors:
Tom Knight; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The lack of standardization in assembly techniques for DNA sequences forces each DNA assembly reaction to be both an experimental tool for addressing the current research topic, and an experiment in and of itself. One of our goals is to replace this ad hoc experimental design with a set of standard and reliable engineering mechanisms to remove much of the tedium and surprise during assembly of genetic components into larger ... |
|
| Extracting Structural Paraphrases from Aligned Monolingual Corpora |
2003 |
9 pages |
| Authors:
Ali Ibrahim; Boris Katz; Jimmy Lin; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| We present an approach for automatically learning paraphrases from aligned monolingual corpora. Our algorithm works by generalizing the syntactic paths between corresponding anchors in aligned sentence pairs. Compared to previous work, structural paraphrases generated by our algorithm tend to be much longer on average, and are capable of capturing long-distance dependencies. In addition to a standalone evaluation of our paraphrases, we also describe a question answering application currently under development ... |
|
| Ensuring Survivable Information Services: Architecting, Diagnosing and Reconfiguring |
OCT 2002 |
36 pages |
| Authors:
Howard Shrobe; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The central premise of this work is that a significant shift of focus and approach is necessary in order to deal with the emerging threat of information attacks on the critical infrastructures of our society. A range of studies, for example those conducted by the RAND Corporation and by a DARPA ISAT project, raised the concern that a concerted information attack by a skillful and motivated opponent could lead to ... |
|
| Exact Solution of the Nonlinear Dynamics of Recurrent Neural Mechanisms for Direction Selectivity |
AUG 2002 |
8 pages |
| Authors:
M. A. Giese; X. Xie; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Different theoretical models have been used to investigate the feasibility of recurrent neural mechanisms for achieving direction selectivity in the visual cortex. The mathematical analysis of such models has been restricted so far to the case of purely linear networks. In this paper, the authors present an exact analytical solution of the nonlinear dynamics of a class of direction selective recurrent neural models with threshold nonlinearity. The mathematical analysis shows ... |
|
| Biologically Plausible Neural Model for the Recognition of Biological Motion and Actions |
AUG 2002 |
27 pages |
| Authors:
Martin A. Giese; Tomaso Poggio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The visual recognition of complex movements and actions is crucial for communications and survival in many species. Remarkable sensitivity and robustness of biological motion perception have been demonstrated in psychophysical experiments. In recent years, neurons and cortical areas involved in action recognition have been identified in neurophysiological and imaging studies. However, the detailed neural mechanisms that underlie the recognition of such complex movement patterns remain largely unknown. This paper reviews ... |
|
| Modeling Stock Order Flows and Learning Market-Making from Data |
JUN 2002 |
9 pages |
| Authors:
Adlar J. Kim; Christian R. Shelton; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Stock markets employ specialized traders, market-makers, designed to provide liquidity and volume to the market by constantly supplying both supply and demand. In this paper, we demonstrate a novel method for modeling the market as a dynamic system and a reinforcement learning algorithm that learns profitable market-making strategies when run on this model. The sequence of buys and sells for a particular stock, the order flow, we model as an ... |
|
| DARPA BioComp Plasmid Distribution 1.00 of Standard Biobrick Components |
22 MAY 2002 |
3 pages |
| Authors:
Tom Knight; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This distribution consists of 300 ng aliquots of plasmid DNA for each of twelve components and compound constructs utilizing our idempotent assembly strategy. The dried DNA should be stable at room temperature for many weeks, but long term storage at -20 or -80 in the resealable foil pouch with desiccant is recommended. Much of the power of these assembly techniques arise from a consistent, widely available set of components. We ... |
|
| Bagging Regularizes |
MAR 2002 |
9 pages |
| Authors:
Tomaso Poggio; Ryan Rifkin; Sayan Mukherjee; Alex Rakhlin; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Intuitively, we expect that averaging - or bagging - different regressors with low correlation should smooth their behavior and be somewhat similar to regularization. In this note we make this intuition precise. Using an almost classical definition of stability, we prove that a certain form of averaging provides generalization bounds with a rate of convergence of the same order as Tikhonov regularization - similar to fashionable RKHS-based learning algorithms, |
|
| Better Vision Through Manipulation |
2002 |
|
| Authors:
Giorgio Metta; Paul Fitzpatrick; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
 | For the purposes of manipulation, we would like to know what parts of the environment are physically coherent ensembles -- that is, which parts will move together, and which are more or less independent. It takes a great deal of experience before this judgement can be made from purely visual information. This paper develops active strategies for acquiring that experience through experimental manipulation, using tight correlations between arm motion and ... |
|
| First Contact: an Active Vision Approach to Segmentation |
2002 |
6 pages |
| Authors:
P. Fitzpatrick; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| How a robot should grasp an object depends on its size and shape. Such parameters can be estimated visually, but this is fallible, particularly for unrecognized, unfamiliar objects. Failure will result in a clumsy grasp or glancing blow against the object. If the robot does not learn something form the encounter, then it will be apt to repeat the same mistake again and again. This paper shows how to recover ... |
|
| Grounding Vision through Experimental Manipulation |
2002 |
21 pages |
| Authors:
Paul Fitzpatrick; Giorgio Metta; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Experimentation is crucial to human progress at all scales, from society as a whole to a young infant in its cradle/ It allows us to elicit learning episodes suited to our own needs and limitations. This paper develops active strategies for a robot to acquire visual experience through simple experimental manipulation. The experiments are oriented towards determining what parts of the environment are physically coherent- that is, which parts will ... |
|
| Behavior-Based Early Language Development on a Humanoid Robot |
2002 |
|
| Authors:
Paulina Varshavskaya; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| We are exploring the idea that early language acquisition could be better modelled on an artificial creature by considering the pragmatic aspect of natural language and of its development in human infants. We have implemented a system of vocal behaviors on Kismet in which "words" or concepts are behaviors in a competitive hierarchy. This paper reports on the framework, the vocal system's architecture and algorithms, and some preliminary results from ... |
|
| A Unified Multiresolution Framework for Automatic Target Recognition (ATR) |
SEP 2001 |
125 pages |
| Authors:
John Fisher; Eric Grimson; Alan Willsky; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This report describes the development of multiscale, multiresolution methods for automatic target recognition (ATR). The methods are applied to and developed specifically for synthetic aperture radar data. Applications include high level reasoning over learned target models, information theoretic approaches for pose estimation, and model development from multiple views. |
|
| Amorphous and Cellular Computing |
31 AUG 2001 |
12 pages |
| Authors:
Harold Abelson; Gerald J. Sussman; Thomas F. Knight Jr; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| The objective of this research is to create the architectural, algorithmic, and technological foundations for exploiting programmable materials. These are materials that incorporate vast numbers of programmable elements that react to each other and to their environment. Such materials can be fabricated economically, provided that the computing elements are amassed in bulk without arranging for precision interconnect and testing. In order to exploit programmable materials we must identify engineering principles ... |
|
| The Intelligent Room |
AUG 2001 |
53 pages |
| Authors:
H. Shrobe; M. Coen; K. Wilson; L. Weisman; K. Thomas; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Intelligent, interactive environments promise to drastically change our everyday lives by connecting computation to ordinary, human-level events happening in the real world. The motivation for building these intelligent, interactive environments is to bring computation into the real, physical world to support what are traditionally considered non-computational activities. The ultimate goal is to allow people to interact with computational systems in a way that they would with other people: via gesture, ... |
|
| From Word-Spotting to OOV Modeling |
2001 |
5 pages |
| Authors:
Paul Fitzpatrick; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| This paper explores one dimension along which word spotting and speech recognition differ: the nature of the background model. In word spotting, a relatively small number of keywords float on a sea of unknown words. In speech recognition, an occasional unknown word punctuates utterances that are otherwise completely within the vocabulary. Despite this difference in viewpoint, in some circumstances implementations of the two may become very similar. When transcribed data ... |
|
| Towards Manipulation-Driven Vision |
2001 |
6 pages |
| Authors:
Paul M. Fitzpatrick; Giorgio Metta; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| For the purposes of manipulation, the authors would like to know what parts of the environment are physically coherent ensembles, that is, which parts will move together, and which parts are more or less independent. It takes a great deal of experience before this judgement can be made from purely visual information. This paper develops active strategies for acquiring that experience through experimental manipulation, using tight correlations between arm motion ... |
|
| Learning about Objects through Action - Initial Steps towards Artificial Cognition |
2001 |
6 pages |
| Authors:
Paul M. Fitzpatrick; Giorgio Metta; Lorenzo Natale; Sajit Rao; Giulio Sandini; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Within the field of Neuro Robotics we are driven primarily by the desire to understand how humans and animals live and grow and solve everyday problems. To this aim we adopted a "learn by doing" approach by building artificial systems, e.g. robots, that not only look like human beings but also represent a model of some brain process. They should, ideally, behave and interact like human beings (being situated). The ... |
|
| Computational Models of Object Recognition in Cortex: A Review |
07 AUG 2000 |
12 pages |
| Authors:
Maximilian Riesenhuber; Tomaso Poggio; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Understanding how biological visual systems perform object recognition is one of the ultimate goals in computational neuroscience. Among the biological models of recognition the main distinctions are between feedforward and feedback and between object-centered and view-centered. From a computational viewpoint the different recognition tasks -- for instance categorization and identification -- are very similar, representing different trade-offs between specificity and invariance. Thus the different tasks do not strictly require different ... |
|
| Regulation and Entrainment in Human-Robot Interaction |
2000 |
18 pages |
| Authors:
Cymthia Breazeal; MASSACHUSETTS INST OF TECH CAMBRIDGE ARTIFICIAL INTELLIGENCE LAB
|
| Newly emerging robotics applications for domestic or entertainment purposes are slowly introducing autonomous robots into society at large. A critical capability of such robots is their ability to interact with humans, and in particular, untrained users. This paper explores the hypothesis that people will intuitively interact with robots in a natural social manner provided the robot can perceive, interpret, and appropriately respond with familiar human social cues. Two experiments are ... |
|
Reports by Corporate Author
