Rapid, Autonomous Learning of Visual Object Categories in Robots and Infants (original) (raw)
Related papers
Human Infant Looking and Learning Behavior in Robots
This paper presents a biologically realistic spiking neural model of the auditory, visual, and looking-control systems of a young human infant. Several behaviors that have been observed in young human infants arise when the model is used to control the head movements of a NAO humanoid robot in the same situations. It demonstrates visual habituation as measured by looking time. It also demonstrates multimodal habituation to synchronous auditory-visual stimuli during real time interaction with a human who is performing a "showing-and-naming" behavior. In-depth analysis of the evolution of the model affords greater understanding of its internal neural dynamics and how they interact with the external world that includes the robot's body and the human. Since the model is biologically based, this increases our understanding of how and why human infants behave as they do, and what can be learned from them to make better interacting robots.
Robots Learn to Recognize Individuals from Imitative Encounters with People and Avatars OPEN
Prior to language, human infants are prolific imitators. Developmental science grounds infant imitation in the neural coding of actions, and highlights the use of imitation for learning from and about people. Here, we used computational modeling and a robot implementation to explore the functional value of action imitation. We report 3 experiments using a mutual imitation task between robots, adults, typically developing children, and children with Autism Spectrum Disorder. We show that a particular learning architecture-specifically one combining artificial neural nets for (i) extraction of visual features, (ii) the robot's motor internal state, (iii) posture recognition, and (iv) novelty detection-is able to learn from an interactive experience involving mutual imitation. This mutual imitation experience allowed the robot to recognize the interactive agent in a subsequent encounter. These experiments using robots as tools for modeling human cognitive development, based on developmental theory, confirm the promise of developmental robotics. Additionally, findings illustrate how person recognition may emerge through imitative experience, intercorporeal mapping, and statistical learning. By 18 months of age human infants are able to recognize themselves in a mirror. This skill is rare in the animal kingdom, and shared only with a few other mammals (e.g., great apes and elephants) 1. The ontogenetic factors contributing to this implicit sense of self have been explored 2. Some component skills are the ability of human infants to discriminate between faces 3 , to compare different inputs and match them across different sensory modalities 4 , and to be sensitive to interpersonal synchrony 5,6. These findings and others raise the intriguing possibility that young infants may be able to detect and use the equivalences between felt acts of the self and visible acts of the other 7 prior to language and before they have compared self and other in a mirror. Here, we use computational modeling and robotics to illuminate a key aspect of preverbal social cognition-how infants use social encounters, especially naturally occurring mutual imitation between adult and child, to help recognize individuals when they are reencountered at another point in time. The experiments reported here use a wide range of social agents, including typically developing adults and children as well as children with autism spectrum disorder (ASD), and avatars. It has been demonstrated that in interpersonal interactions preverbal infants do not just recognize that another moves when they move (temporal contingency), but that another acts in the same manner as they do (structural congruence) 6–8. This has been shown by measures of increased attention and positive affect at being imitated, as well as by neuroscience measures acquired during mutual imitation episodes (mu rhythm responses in the infant electroencephalogram, EEG) 9. Such recognitions imply that there is a coding of one's own body and its relation to the body of others prior to language, and raises the idea that preverbal action imitation is a mechanism for social learning as sought by evolutionary biologists 10 , and a channel for preverbal communication 11. Within developmental psychology, the recognition of personal identity is thought to be a crucial developmental milestone, because infants need to re-identify a person as " the same one again " after a break in perceptual contact and after changes in appearance (putting on a kerchief, growing a beard, getting a haircut) 12. Once the child has language,
We present a perceptual system for a humanoid robot that autonomously discovers its hand from less than 5 minutes of natural interaction with a human. The perceptual system combines simple proprioceptive sensing with a visual attention system that uses motion to select salient regions. We show that during natural interactions with a person, the selected visual regions primarily correspond with the person's head, eyes, hands, and fingers, and the robot's hand and fingers. The system clusters the selected image regions, models their spatial distribution over a sensory ego-sphere, and uses mutual information to determine how well the robot's arm can control each cluster. In our tests, the cluster that is best controlled by the robot's arm primarily contains images of the robot's hand. It also has a spatial distribution that can predict the visual location of the robot's hand as a function of the arm's configuration. 1
Computer Vision for Learning to Interact Socially with Humans
Concepts, Methodologies, Tools, and Applications
Computer vision is essential to develop a social robotic system capable to interact with humans. It is responsible to extract and represent the information around the robot. Furthermore, a learning mechanism, to select correctly an action to be executed in the environment, pro-active mechanism, to engage in an interaction, and voice mechanism, are indispensable to develop a social robot. All these mechanisms together provide a robot emulate some human behavior, like shared attention. Then, this chapter presents a robotic architecture that is composed with such mechanisms to make possible interactions between a robotic head with a caregiver, through of the shared attention learning with identification of some objects.
Infant-like Social Interactions between a Robot and a Human Caregiver
Adaptive Behavior, 2000
This paper presents an autonomous robot designed to interact socially with human "parents". A human infant's emotions and drives play an important role in generating meaningful interactions with the caretaker, regulating these interactions to maintain an environment suitable for the learning process, and assisting the caretaker in satisfying the infant's drives. For our purposes, the ability to regulate how intensely the caretaker engages the robot is vital to successful learning in a social context.
Perception and human interaction for developmental learning of objects and affordances
In this paper we describe a cognitive architecture for humanoids interacting with objects and caregivers in a developmental robotics scenario. The architecture is foundational to the MACSi project: it is designed to support experiments to make a humanoid robot gradually enlarge its repertoire of known objects and skills combining autonomous learning, social guidance and intrinsic motivation. This complex learning process requires the capability to learn affordances first. Here, we present the general framework for achieving these goals, focusing on the elementary action, perception and interaction modules. Preliminary experiments performed on the humanoid robot iCub are also discussed.
Robots Learn to Recognize Individuals from Imitative Encounters with People and Avatars
Prior to language, human infants are prolific imitators. Developmental science grounds infant imitation in the neural coding of actions, and highlights the use of imitation for learning from and about people. Here, we used computational modeling and a robot implementation to explore the functional value of action imitation. We report 3 experiments using a mutual imitation task between robots, adults, typically developing children, and children with Autism Spectrum Disorder. We show that a particular learning architecture-specifically one combining artificial neural nets for (i) extraction of visual features, (ii) the robot's motor internal state, (iii) posture recognition, and (iv) novelty detection-is able to learn from an interactive experience involving mutual imitation. This mutual imitation experience allowed the robot to recognize the interactive agent in a subsequent encounter. These experiments using robots as tools for modeling human cognitive development, based on developmental theory, confirm the promise of developmental robotics. Additionally, findings illustrate how person recognition may emerge through imitative experience, intercorporeal mapping, and statistical learning. By 18 months of age human infants are able to recognize themselves in a mirror. This skill is rare in the animal kingdom, and shared only with a few other mammals (e.g., great apes and elephants) 1. The ontogenetic factors contributing to this implicit sense of self have been explored 2. Some component skills are the ability of human infants to discriminate between faces 3 , to compare different inputs and match them across different sensory modalities 4 , and to be sensitive to interpersonal synchrony 5,6. These findings and others raise the intriguing possibility that young infants may be able to detect and use the equivalences between felt acts of the self and visible acts of the other 7 prior to language and before they have compared self and other in a mirror. Here, we use computational modeling and robotics to illuminate a key aspect of preverbal social cognition-how infants use social encounters, especially naturally occurring mutual imitation between adult and child, to help recognize individuals when they are reencountered at another point in time. The experiments reported here use a wide range of social agents, including typically developing adults and children as well as children with autism spectrum disorder (ASD), and avatars. It has been demonstrated that in interpersonal interactions preverbal infants do not just recognize that another moves when they move (temporal contingency), but that another acts in the same manner as they do (structural congruence) 6–8. This has been shown by measures of increased attention and positive affect at being imitated, as well as by neuroscience measures acquired during mutual imitation episodes (mu rhythm responses in the infant electroencephalogram, EEG) 9. Such recognitions imply that there is a coding of one's own body and its relation to the body of others prior to language, and raises the idea that preverbal action imitation is a mechanism for social learning as sought by evolutionary biologists 10 , and a channel for preverbal communication 11. Within developmental psychology, the recognition of personal identity is thought to be a crucial developmental milestone, because infants need to re-identify a person as " the same one again " after a break in perceptual contact and after changes in appearance (putting on a kerchief, growing a beard, getting a haircut) 12. Once the child has language,
“Social” robots are psychological agents for infants: A test of gaze following
2010
Gaze following is a key component of human social cognition. Gaze following directs attention to areas of high information value and accelerates social, causal, and cultural learning. An issue for both robotic and infant learning is whose gaze to follow. The hypothesis tested in this study is that infants use information derived from an entity’s interactions with other agents as evidence about whether that entity is a perceiver. A robot was programmed so that it could engage in communicative, imitative exchanges with an adult experimenter. Infants who saw the robot act in this social-communicative fashion were more likely to follow its line of regard than those without such experience. Infants use prior experience with the robot’s interactions as evidence that the robot is a psychological agent that can see. Infants want to look at what the robot is seeing, and thus shift their visual attention to the external target.
Infants' object processing is guided specifically by social cues
Neuropsychologia, 2019
Previous studies showed that the movements of another person's eyes and head guides infants' attention and promotes social learning by leading to enhanced encoding of cued objects. However, it is an open question whether social features like eyes are required or if the lateral movement of any arbitrary stimulus can elicit similar effects. The current experiments investigate the effects of the movement of a nonsocial cue and a perceptually similar social cue on object processing in 4-month-olds using event-related potentials (ERPs). Infants were presented with one of two central cues, either a box with a checkerboard pattern or a box with eye-like features on the front, which turned to one side. The cue thereby either turned toward a novel object or turned away from it. Afterwards, the object was presented again and ERPs in response to these previously cued or uncued objects were compared. When the nonsocial box served as the cue, no difference in neural processing of previously cued and uncued objects was found. In contrast, when the box with eyes served as the cue, we found an enhanced positive slow wave (PSW) for uncued as compared to cued objects. While the turning of the box with eyes promoted the encoding of cued objects, uncued objects needed enhanced activity for processing when presented for a second time. Results suggest that not every dynamic cue can influence infants' object processing but that the presence of a basic social characteristic like isolated schematic eyes is sufficient to enhance social learning processes in early infancy. This hints on a specific sensitivity of the infant brain to social information which helps infants to focus on relevant information in the environment during social learning.