- Agents with Anticipatory Behaviors: To be Cautious in a Risky Environment 2007-03-13
Cristiano Castelfranchi and Rino Falcone and Michele Piunti, European Conference Articial Intelligence (ECAI06) pp. 693-694
- “Exhibitionists” and “Voyeurs” do it better: a Shared Environment for Flexible Coordination with Tacit Messages 2007-04-10
Tummolini, L., Castelfranchi C., Ricci, A. Viroli, M., Omicini, A. (2005); in H. van Parunak, D. Weyns (Eds.) Lecture Notes in Artificial Intelligence, LNAI 3374, Sringer-Verlag Heidelberg, pp. 215-231.
- Knowing the unknown 2013-03-27
- Robot butlers that tidy your house or cook you a meal have long been the dream of science-fiction writers and artificial intelligence researchers alike.
But if robots are ever going to move effectively around our constantly changing homes or workspaces performing such complex tasks, they will need to be more aware of their own limitations, according to researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL).
Most successful robots today tend to be used either in fixed, carefully controlled environments, such as manufacturing plants, or for performing fairly simple tasks such as vacuuming a room, says Leslie Pack Kaelbling, the Panasonic Professor of Computer Science and Engineering at MIT.
Carrying out complicated sequences of actions in a cluttered, dynamic environment such as a home will require robots to be more aware of what they do not know, and therefore need to find out, Kaelbling says. That is because a robot cannot simply look around the kitchen and determine where all the containers are stored, for example, or what you would prefer to eat for dinner. To find these things out, it needs to open the cupboards and look inside, or ask a question.
“I would like to make a robot that could go into your kitchen for the first time, having been in other kitchens before but not yours, and put the groceries away,” Kaelbling says.
And in a paper recently accepted for publication in the International Journal of Robotics Research, she and CSAIL colleague Tomas Lozano-Perez describe a system designed to do just that, by constantly calculating the robot’s level of uncertainty about a given task, such as the whereabouts of an object, or its own location within the room.
The system is based on a module called the state estimation component, which calculates the probability of any given object being what or where the robot thinks it is. In this way, if the robot is not sufficiently certain that an object is the one it is looking for, because the probability of it being that object is too low, it knows it needs to gather more information before taking any action, Kaelbling says.
So, for example, if the robot were trying to pick up a box of cereal from a shelf, it might decide its uncertainty about the position of the object was too high to attempt grasping it. Instead, it would first take a closer look at the object, in order to get a better idea of its exact location, Kaelbling says. “It’s thinking always about its own belief about the world, and how to change its belief, by taking actions that will either gather more information or change the state of the world.”
The system also simplifies the process of developing a strategy for performing a given task by making up its plan in stages as it goes along, using what the team calls hierarchical planning in the now.
“There is this idea in AI that we’re very worried about having an optimal plan, so we’re going to compute very hard for a long time, to ensure we have a complete strategy formulated before we begin execution,” Kaelbling says.
But in many cases, particularly if the environment is new to the robot, it cannot know enough about the area to make such a detailed plan in advance, she says.
So instead the system makes a plan for the first stage of its task and begins executing this before it has come up with a strategy for the rest of the exercise. That means that instead of one big complicated strategy, which consumes a considerable amount of computing power and time, the robot can make many smaller plans as it goes along.
The drawback to this process is that it can lead the robot into making silly mistakes, such as picking up a plate and moving it over to the table without realizing that it first needs to clear some room to put it down, Kaelbling says.
But such small mistakes may be a price worth paying for more capable robots, she says: “As we try to get robots to do bigger and more complicated things in more variable environments, we will have to settle for some amount of suboptimality.”
In addition to household robots, the system could also be used to build more flexible industrial devices, or in disaster relief, Kaelbling says.
Ronald Parr, an associate professor of computer science at Duke University, says much existing work on robot planning tends to be fragmented into different groups working on particular, specialized problems. In contrast, the work of Kaelbling and Lozano-Perez breaks down the walls that exist between these subgroups, and uses hierarchical planning to address the computational challenges that arise when attempting to develop a more general-purpose, problem-solving system. “What’s more, it is demonstrated on a practical, general-purpose robotic platform that could be used for domestic or factory work,” Parr says.
- Rappresentazioni Anticipatorie: Tre Studi Simulativi 2007-04-15
Giovanni Pezzulo (2006) Proceedings del 3° Convegno Nazionale di Scienze Cognitive (AISC 2006).
- Brain’s language center has multiple roles 2012-10-16
- A century and a half ago, French physician Pierre Paul Broca found that patients with damage to part of the brain’s frontal lobe were unable to speak more than a few words. Later dubbed Broca’s area, this region is believed to be critical for speech production and some aspects of language comprehension.
However, in recent years neuroscientists have observed activity in Broca’s area when people perform cognitive tasks that have nothing to do with language, such as solving math problems or holding information in working memory. Those findings have stimulated debate over whether Broca’s area is specific to language or plays a more general role in cognition.
A new study from MIT may help resolve this longstanding question. The researchers, led by Nancy Kanwisher, the Walter A. Rosenblith Professor of Cognitive Neuroscience, found that Broca’s area actually consists of two distinct subunits. One of these focuses selectively on language processing, while the other is part of a brainwide network that appears to act as a central processing unit for general cognitive functions.
“I think we’ve shown pretty convincingly that there are two distinct bits that we should not be treating as a single region, and perhaps we shouldn’t even be talking about ‘Broca’s area’ because it’s not a functional unit,” says Evelina Fedorenko, a research scientist in Kanwisher’s lab and lead author of the new study, which recently appeared in the journal Current Biology.
Kanwisher and Fedorenko are members of MIT’s Department of Brain and Cognitive Sciences and the McGovern Institute for Brain Research. John Duncan, a professor of neuroscience at the Cognition and Brain Sciences Unit of the Medical Research Council in the United Kingdom, is also an author of the paper.
A general role
Broca’s area is located in the left inferior frontal cortex, above and behind the left eye. For this study, the researchers set out to pinpoint the functions of distinct sections of Broca’s area by scanning subjects with functional magnetic resonance imaging (fMRI) as they performed a variety of cognitive tasks.
To locate language-selective areas, the researchers asked subjects to read either meaningful sentences or sequences of nonwords. A subset of Broca’s area lit up much more when the subjects processed meaningful sentences than when they had to interpret nonwords.
The researchers then measured brain activity as the subjects performed easy and difficult versions of general cognitive tasks, such as doing a math problem or holding a set of locations in memory. Parts of Broca’s area lit up during the more demanding versions of those tasks. Critically, however, these regions were spatially distinct from the regions involved in the language task.
These data allowed the researchers to map, for each subject, two distinct regions of Broca’s area — one selectively involved in language, the other involved in responding to many demanding cognitive tasks. The general region surrounds the language region, but the exact shapes and locations of the borders between the two vary from person to person.
The general-function region of Broca’s area appears to be part of a larger network sometimes called the multiple demand network, which is active when the brain is tackling a challenging task that requires a great deal of focus. This network is distributed across frontal and parietal lobes in both hemispheres of the brain, and all of its components appear to communicate with one another. The language-selective section of Broca’s area also appears to be part of a larger network devoted to language processing, spread throughout the brain’s left hemisphere.
The findings provide evidence that Broca’s area should not be considered to have uniform functionality, says Peter Hagoort, a professor of cognitive neuroscience at Radboud University Nijmegen in the Netherlands. Hagoort, who was not involved in this study, adds that more work is needed to determine whether the language-selective areas might also be involved in any other aspects of cognitive function. “For instance, the language-selective region might play a role in the perception of music, which was not tested in the current study,” he says.
The researchers are now trying to determine how the components of the language network and the multiple demand network communicate internally, and how the two networks communicate with each other. They also hope to further investigate the functions of the two components of Broca’s area.
“In future studies, we should examine those subregions separately and try to characterize them in terms of their contribution to various language processes and other cognitive processes,” Fedorenko says.
The team is also working with scientists at Massachusetts General Hospital to study patients with a form of neurodegeneration that gradually causes loss of the ability to speak and understand language. This disorder, known as primary progressive aphasia, appears to selectively target the language-selective network, including the language component of Broca’s area.
The research was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the Ellison Medical Foundation and the U.K. Medical Research Council.
- Expectations driven approach for Situated, Goal-directed Agents 2007-09-26
In proceedings of AI*IA/TABOO Joint Workshop (WOA 2007). Genova, Italy. 2007
- Learning to anticipate a temporarily hidden moving object 2007-06-01
- McGovern Institute hands out neuroscience award 2013-02-28
- The McGovern Institute for Brain Research at MIT has announced that Thomas Jessell of Columbia University is the winner of the 2013 Edward M. Scolnick Prize in Neuroscience. The $100,000 prize, which is endowed through a gift from Merck, is awarded annually by the McGovern Institute to recognize outstanding advances in the field of neuroscience.
Jessell received the award for his research on the embryonic development of the nervous system. His primary interest is the development of the spinal cord, which because of its relative simplicity and evolutionary conservation offers an ideal system for understanding general principles of central nervous system development.
Jessell’s work has revealed the molecular mechanisms responsible for establishing the spatial organization of the spinal cord. He has identified diffusible signaling molecules that act during early development to provide "positional information" to embryonic cells, instructing them to differentiate in ways that are appropriate for their specific locations within the cord. Jessell has also studied the molecular mechanisms by which developing cells respond to these positional cues. This work has led to the identification of a transcriptional code, whereby a set of regulatory genes act in combination to specify the many different cell types that comprise the mature spinal cord.
The discovery of these genetic mechanisms has made it possible to identify and manipulate the activity of specific classes of neurons with great precision, and Jessell has used this approach to reveal the link between functional circuitry and motor behavior.
In addition to fundamental questions, Jessell’s work has important practical implications for the emerging field of regenerative medicine. There is great interest in stem cells as a renewable source of cells for transplantation therapy and drug discovery, but for this approach to succeed, stem cells must be converted to the desired cell type. Jessell’s work on transcriptional control of neural identity provides a roadmap for such efforts, and he has demonstrated its feasibility by converting embryonic stem cells into spinal motor neurons, the same cell types that degenerate in diseases such as amyotrophic lateral sclerosis. The implications of his research go well beyond motor neuron diseases; many disorders of the nervous system affect particular cell types, and the ability to convert stem cells to specific classes of neurons may eventually find wide applications in clinical neuroscience.
The McGovern Institute will award the Scolnick Prize to Dr. Jessell on Monday, April 1, 2013. He will deliver a lecture entitled “Sifting Circuits for Motor Control" at 4 p.m. that day, to be followed by a reception, at the McGovern Institute in the Brain and Cognitive Sciences Complex, 43 Vassar Street (building 46, room 3002). The event is free and open to the public.
- Apprendimento per rinforzo e codifica tramite popolazione neurale: un modello per il reaching applicato a due task 2007-03-19
WIVA3 - 3° Workshop Italiano Vita Artificiale