Abstract not found

An increasing amount of evidence suggests that in human infants the ability to learn by watching others, and in particular, the ability to imitate, could be crucial precursors

The processes and representations used to generate the behavior of expressive virtual characters are a valuable and largely untapped resource for helping those characters make sense of the world around them. In this paper, we present Max T. Mouse, an anthropomorphic animated mouse character who uses his own motor and behavior representations to interpret the behaviors he sees his friend Morris Mouse performing. Specifically, by using his own motor and action systems as models for the behavioral capabilities of others (a process known as Simulation Theory in the cognitive literature), Max can begin to identify simple goals and motivations for Morris’s behavior, an important step towards developing socially intelligent animated characters. Additionally, Max uses a novel motion graph-based movement recognition process in order to accurately parse and imitate Morris’s movements and behaviors in real-time and without prior examples, even when provided with limited synthetic visual input. Key contributions of this paper include demonstrating that using the same mechanisms for movement and behavior perception and production allows for an elegant conservation of representation, and that the innate structure of motion graphs can be used to facilitate both movement parsing and movement recognition.

We want to build humanoid robots capable of rich social interactions with humans, including natural communication, cooperation, and tutelage. Humans display a remarkably flexible and rich array of social competencies, demonstrating the ability to interpret, predict, and respond appropriately to the behavior of others. People understand each other in social terms, allowing us to engage others in a variety of complex social interactions including communication, social learning, and cooperation. Developing robots with these social abilities is a critical step towards enabling them to be intelligent and capable in their interactions with humans, able to cooperate with people as capable partners, able to learn quickly and effectively from natural human instruction, are intuitive to communicate with, and are engaging for humans to interact with. Such issues must be addressed to enable many new and exciting applications for robots that require them to play a long-term role in people's daily lives. This paper presents an overview of our work towards building socially intelligent, cooperative humanoid robots that can work and learn in partnership with people.

Imitation o#ers a powerful mechanism for knowledge acquisition, particularly for intelligent agents (like infants) that lack the ability to transfer knowledge using language. Several algorithms and models have recently been proposed for imitation learning in humans and robots. However, few proposals o#er a framework for imitation learning in noisy stochastic environments where the imitator must learn and act under real-time performance constraints. In this paper, we present a novel probabilistic framework for imitation learning in stochastic environments with unreliable sensors. We develop Bayesian algorithms, based on Meltzo# and Moore's AIM hypothesis for infant imitation, that implement the core of an imitation learning framework. Our algorithms are computationally e#cient, allowing real-time learning and imitation in an active stereo vision robotic head. We present results of both software simulations and our algorithms running on the head, demonstrating the validity of our approach. We conclude by advocating a research agenda that Preprint submitted to Elsevier Science 4 October 2004 promotes interaction between cognitive and robotic studies of imitation.

As robots enter the human environment they must be able to communicate and cooperate with novice users. Towards this goal, understanding human nonverbal behavior is a critical skill. This includes not only recognizing human’s actions, but also inferring mental states from observable behavior. This capability would allow a robot to offer predictive and relevant assistance to a human. Simulation Theory argues in favor of an embodied approach for how humans infer mental states of others (e.g., intents, beliefs, affect, etc.). This theory argues that humans reuse parts of their cognitive structure used for generating behavior to simulate and detect the behavior of others. Inspired by this theory, we describe our simulation-theoretic approach and implementation that enables a robot to monitor the human by simulating their behavior within the robot’s own generative mechanisms on the motion, action, and perceptual levels. This grounds the robot’s information about the user in the robot’s own systems, allowing it to make inferences about the human’s goals and knowledge that are immediately useful for providing helpful behavior such as helping to complete an action or pointing out an occluded object. We feel that designing individual systems of the robot to allow for this type of dual use, and reusing them in this manner, is a powerful technique for designing robots that interact with humans.

observation of humanoid robot movements elicits

Robot programming by demonstration (PbD) has become a central topic of robotics that spans across general research areas such as humanrobot interaction, machine learning, machine vision and motor control. Robot PbD started about 30 years ago, and has grown importantly during the past decade. The rationale for moving from purely preprogrammed robots to very flexible user-based interfaces for training robots to perform a task is three-fold. First and foremost, PbD, also referred to as imitation learning, is a powerful mechanism for reducing the complexity of search spaces for learning. When observing either good or bad examples, one can reduce the search for a possible solution,

We developed an assistive robot for children with autism in order to enhance their primal imitation skills. Our approach directed toward encouraging autistic children to engage in physical exercise during a morning gathering through imitating an embedded robotic system. Through these interaction scenarios we attempt to heighten the children’s imitation skills through a music-embedded imitation interaction with a robot. The most significant part of this project was the development of a robot imitation algorithm to imitate a child’s motions. Accordingly, we propose the use of a single camera-based image-processing technique with a novel imitation algorithm for imitating a child’s motions. The results of this study show that proposed imitation algorithm was able to accurately imitate the target human motions. 1.

Modeling human behavior is important for the design of robots as well as humancomputer interfaces that use humanoid avatars. Constructive models have been built, but they have not captured all of the detailed structure of human behavior such as the moment-to-moment deployment and coordination of hand, head and eye gaze used in complex tasks. We show how this data from human subjects performing a task can be used to program a dynamic Bayes network (DBN) which in turn can be used to recognize new performance instances. As a specific demonstration we show that the steps in a complex activity such as sandwich making can be recognized by a DBN in real time.