Paul Benjamin - Academia.edu (original) (raw)

Papers by Paul Benjamin

International Joint Conference on Artificial Intelligence, Aug 23, 1987

One of the major 'weaknesses of current automated reasoning systems is that they lack the ability... more One of the major 'weaknesses of current automated reasoning systems is that they lack the ability to con trol inference in a sophisticated, context-directed fashion. General strategies such as the set-of-support strategy are useful, but have proven inadequate for many individual problems. A strategy component is needed that possesses knowledge about many particu lar domains and problems. Such a body of knowledge would require a prohibitive amount of time to con struct by hand. This leads us to consider means of automatically acquiring control knowledge from exam ple proofs. One particular means of learning is explanation-based learning. This paper analyzes the basis of explanations -finding weakest preconditions that enable a particular rule to fire -to derive a representation within which explanations can be extracted from examples, generalized and used to guide the actions of a problem-solving system.

arXiv (Cornell University), Dec 10, 2014

Recognition of objects using Deep Neural Networks is an active area of research and many breakthr... more Recognition of objects using Deep Neural Networks is an active area of research and many breakthroughs have been made in the last few years. The paper attempts to indicate how far this field has progressed. The paper briefly describes the history of research in Neural Networks and describe several of the recent advances in this field. The performances of recently developed Neural Network Algorithm over benchmark datasets have been tabulated. Finally, some the applications of this field have been provided.

ACM SIGCSE Bulletin, 2003

Mathematics and Computers in Simulation, 1990

ABSTRACT The ADAPT project is a collaboration of researchers in robotics, linguistics and artific... more ABSTRACT The ADAPT project is a collaboration of researchers in robotics, linguistics and artificial intelligence at three universities to create a cognitive architecture specifically designed to be embodied in a mobile robot. There are major respects in which existing cognitive architectures are inadequate for robot cognition. In particular, they lack support for true concurrency and for active perception. ADAPT addresses these deficiencies by modeling the world as a network of concurrent schemas, and modeling perception as problem solving. Schemas are represented using the RS (Robot Schemas) language, and are activated by spreading activation. RS provides a powerful language for distributed control of concurrent processes. Also, The formal semantics of RS provides the basis for the semantics of ADAPT's use of natural language. We have implemented the RS language in Soar, a mature cognitive architecture originally developed at CMU and used at a number of universities and companies. Soar's subgoaling and learning capabilities enable ADAPT to manage the complexity of its environment and to learn new schemas from experience. We describe the issues faced in developing an embodied cognitive architecture, and our implementation choices.

Procedia Manufacturing, 2015

There is a growing body of evidence that human perception is "active", in the sense that it is la... more There is a growing body of evidence that human perception is "active", in the sense that it is largely goal-oriented and top-down. Task goals appear to influence how people perceive their environment. Effective interaction between people and unmanned systems requires that the unmanned systems' perceptions be comprehensible to people, and this means that the unmanned systems should also perceive the world in an active manner. Recent evidence in cognitive psychology and neuroscience supports the proposition that simulation, the "re-enactment of perceptual, motor and introspective states" is a central cognitive mechanism. Cognitive functions such as anticipation and planning operate through a process of internal simulation of actions and environment. Indeed there is a history in the field of Artificial Intelligence of using "simulated action" as an algorithmic search procedure, e.g., game trees, though such an approach typically has problematic computational complexity. The simulations include not just the effect of actions, but also the understood laws of physics (e.g., will a falling object continue to fall). We are building a robot cognitive architecture that is based on a unified cognitive architecture-Soar-and that uses active perception and simulation in planning. Our system constructs a 3D virtual copy of itself and its environment, including people, and updates this model in realtime to agree with changes in its environment. This 3D model is a qualitative description of the world, and includes physics simulation capabilities so that effects of actions can be simulated before being executed in the real world. This allows the unmanned system to predict and plan its interactions with people. This paper describes the structure of this architecture, and provides examples and videos showing its performance in dynamic environments.

2007 IEEE Symposium on Computational Intelligence in Security and Defense Applications, 2007

The VMSoar project at Pace University is building a cognitive agent for cybersecurity. The projec... more The VMSoar project at Pace University is building a cognitive agent for cybersecurity. The project's objective is to create an intelligent agent that can model and understand the activities of users who are on the network, and that can communicate with network administrators in English to alert them to illegal or suspicious activities. VMSoar can understand users' activities because it is capable of performing these activities itself. It knows how to perform both legal and illegal activities, and uses this knowledge to explore simulations of the activity on a network. It can also probe information stored on a machine to assess the legality of past activity. Research in cybersecurity is difficult is due to the extremely large amount of data that must be analyzed to detect illegal activities. In addition, new exploits are developed frequently. Most current projects in this area are attempting to build some level of intelligence into their systems; however, those projects are focusing primarily on statistical data mining approaches. The VMSoar project is unique in its approach to building an intelligent security agent. The VMSoar agent is based on Soar, a mature cognitive architecture that is used in universities and corporations around the world. I.

We are developing a robotic cognitive architecture to be embedded in autonomous robots that can s... more We are developing a robotic cognitive architecture to be embedded in autonomous robots that can safely interact and collaborate with people on a wide range of physical tasks. Achieving true autonomy requires increasing the robot’s understanding of the dynamics of its world (physical understanding), and particularly the actions of people (cognitive understanding). Our system’s cognitive understanding arises from the Soar cognitive architecture, which constitutes the reasoning and planning component. The system’s physical understanding stems from its central representation, which is a 3D virtual world that the architecture synchronizes with the environment in real time. The virtual world provides a common representation between the robot and humans, thus improving trust between them and promoting effective collaboration.

SPIE Proceedings, 2016

We are building a cognitive vision system for mobile robots that works in a manner similar to the... more We are building a cognitive vision system for mobile robots that works in a manner similar to the human vision system, using saccadic, vergence and pursuit movements to extract information from visual input. At each fixation, the system builds a 3D model of a small region, combining information about distance, shape, texture and motion to create a local dynamic spatial model. These local 3D models are composed to create an overall 3D model of the robot and its environment. This approach turns the computer vision problem into a search problem whose goal is the acquisition of sufficient spatial understanding for the robot to succeed at its tasks. The research hypothesis of this work is that the movements of the robot's cameras are only those that are necessary to build a sufficiently accurate world model for the robot's current goals. For example, if the goal is to navigate through a room, the model needs to contain any obstacles that would be encountered, giving their approximate positions and sizes. Other information does not need to be rendered into the virtual world, so this approach trades model accuracy for speed.

Advances in Intelligent Systems and Computing, 2016

Robots interacting and collaborating with people need to comprehend and predict their movements. ... more Robots interacting and collaborating with people need to comprehend and predict their movements. We present an approach to perceiving and modeling behaviors using a 3D virtual world. The robot's visual data is registered with the virtual world to construct a model of the dynamics of the behavior and to predict future motions using a physics engine. This enables the robot to visualize alternative evolutions of the dynamics and to classify them. The goal of this work is to use this ability to interact more naturally with humans and to avoid potentially disastrous mistakes.

The effective reuse of domain theories in problem solving requires the problem-solving agent to i... more The effective reuse of domain theories in problem solving requires the problem-solving agent to identify general theories whose properties "scale up": they hold for a class of problems of varying size. Otherwise, the agent will be overwhelmed by the cost of indexing and retrieving a huge collection of domain theories, each of which applies in very restricted cases. Furthermore, these general theories need to be represented in a manner that is as independent as possible of the circumstances of particular cases. This paper describes research on analysis and reformulation of domain theories. The perspective of this work is to view a problem space as though it were physical space, and the actions in the problem space as though they were physical motions. A domain theory should then state the laws of motion within the space. Following the analogy with physics, a representation is a coordinate system, and theories are reformulated by transforming coordinates. The mathematical basis for this analogy is briefly given, and illustrated on two simple examples.

Procedia Manufacturing, 2015

There is a growing body of evidence that human perception is "active", in the sense that it is la... more There is a growing body of evidence that human perception is "active", in the sense that it is largely goal-oriented and top-down. Task goals appear to influence how people perceive their environment. Effective interaction between people and unmanned systems requires that the unmanned systems' perceptions be comprehensible to people, and this means that the unmanned systems should also perceive the world in an active manner. Recent evidence in cognitive psychology and neuroscience supports the proposition that simulation, the "re-enactment of perceptual, motor and introspective states" is a central cognitive mechanism. Cognitive functions such as anticipation and planning operate through a process of internal simulation of actions and environment. Indeed there is a history in the field of Artificial Intelligence of using "simulated action" as an algorithmic search procedure, e.g., game trees, though such an approach typically has problematic computational complexity. The simulations include not just the effect of actions, but also the understood laws of physics (e.g., will a falling object continue to fall). We are building a robot cognitive architecture that is based on a unified cognitive architecture-Soar-and that uses active perception and simulation in planning. Our system constructs a 3D virtual copy of itself and its environment, including people, and updates this model in realtime to agree with changes in its environment. This 3D model is a qualitative description of the world, and includes physics simulation capabilities so that effects of actions can be simulated before being executed in the real world. This allows the unmanned system to predict and plan its interactions with people. This paper describes the structure of this architecture, and provides examples and videos showing its performance in dynamic environments.

Multisensor, Multisource Information Fusion: Architectures, Algorithms, and Applications 2015, 2015

ABSTRACT In previous work, we have shown how a 3D model can be built in real time and synchronize... more ABSTRACT In previous work, we have shown how a 3D model can be built in real time and synchronized with the environment. This world model permits a robot to predict dyamics in its environment and classify behaviors. In this paper we evaluate the effect of such a 3D model on the accuracy and speed of various computer vision algorithms, including tracking, optical flow and stereo disparity. We report results based on the KITTI database and on our own videos.

SPIE Proceedings, 2012

We consider the scenario where an autonomous platform that is searching or traversing a building ... more We consider the scenario where an autonomous platform that is searching or traversing a building may observe unstable masonry or may need to travel over unstable rubble. A purely behaviour-based system may handle these challenges but produce behaviour that works against long-terms goals such as reaching a victim as quickly as possible. We extend our work on ADAPT, a cognitive robotics architecture that incorporates 3D simulation and image fusion, to allow the robot to predict the behaviour of physical phenomena, such as falling masonry, and take actions consonant with long-term goals. We experimentally evaluate a cognitive only and reactive only approach to traversing a building filled with various numbers of challenges and compare their performance. The reactive only approach succeeds only 38% of the time, while the cognitive only approach succeeds 100% of the time. While the cognitive only approach produces very impressive behaviour, our results indicate how much better the combination of cognitive and behaviour-based can be.

SPIE Proceedings, 2011

An important component of cognitive robotics is the ability to mentally simulate physical process... more An important component of cognitive robotics is the ability to mentally simulate physical processes and to compare the expected results with the information reported by a robot's sensors. In previous work, we have proposed an approach that integrates a 3D game-engine simulation into the robot control architecture. A key part of that architecture is the Match-Mediated Difference (MMD) operation, an approach to fusing sensory data and synthetic predictions at the image level. The MMD operation insists that simulated and predicted scenes are similar in terms of the appearance of the objects in the scene. This is an overly restrictive constraint on the simulation since parts of the predicted scene may not have been previously viewed by the robot. In this paper we propose an extended MMD operation that relaxes the constraint and allows the real and synthetic scenes to differ in some features but not in (selected) other features. Image difference operations that allow a real image and synthetic image generated from an arbitrarily colored graphical model of a scene to be compared. Scenes with the same content show a zero difference. Scenes with varying foreground objects can be controlled to compare the color, size and shape of the foreground.

Proceedings of the ACM/IEEE international conference on Human-robot interaction, 2007

The ADAPT project is a collaboration of researchers in linguistics, robotics and artificial intel... more The ADAPT project is a collaboration of researchers in linguistics, robotics and artificial intelligence at three universities. We are building a complete robotic cognitive architecture for a mobile robot designed to interact with humans in a range of environments, and which uses natural language and models human behavior. This paper concentrates on the HRI aspects of ADAPT, and especially on how ADAPT models and interacts with humans.

SPIE Proceedings, 2010

This paper describes our work on integrating distributed, concurrent control in a cognitive archi... more This paper describes our work on integrating distributed, concurrent control in a cognitive architecture, and using it to classify perceived behaviors. We are implementing the Robot Schemas (RS) language in Soar. RS is a CSP-type programming language for robotics that controls a hierarchy of concurrently executing schemas. The behavior of every RS schema is defined using port automata. This provides precision to the semantics and also a constructive means of reasoning about the behavior and meaning of schemas. Our implementation uses Soar operators to build, instantiate and connect port automata as needed. Our approach is to use comprehension through generation (similar to NLSoar) to search for ways to construct port automata that model perceived behaviors. The generality of RS permits us to model dynamic, concurrent behaviors. A virtual world (Ogre) is used to test the accuracy of these automata. Soar's chunking mechanism is used to generalize and save these automata. In this way, the robot learns to recognize new behaviors.

2008 ECSIS Symposium on Learning and Adaptive Behaviors for Robotic Systems (LAB-RS), 2008

International Joint Conference on Artificial Intelligence, Aug 23, 1987

One of the major 'weaknesses of current automated reasoning systems is that they lack the ability... more One of the major 'weaknesses of current automated reasoning systems is that they lack the ability to con trol inference in a sophisticated, context-directed fashion. General strategies such as the set-of-support strategy are useful, but have proven inadequate for many individual problems. A strategy component is needed that possesses knowledge about many particu lar domains and problems. Such a body of knowledge would require a prohibitive amount of time to con struct by hand. This leads us to consider means of automatically acquiring control knowledge from exam ple proofs. One particular means of learning is explanation-based learning. This paper analyzes the basis of explanations -finding weakest preconditions that enable a particular rule to fire -to derive a representation within which explanations can be extracted from examples, generalized and used to guide the actions of a problem-solving system.

arXiv (Cornell University), Dec 10, 2014

Recognition of objects using Deep Neural Networks is an active area of research and many breakthr... more Recognition of objects using Deep Neural Networks is an active area of research and many breakthroughs have been made in the last few years. The paper attempts to indicate how far this field has progressed. The paper briefly describes the history of research in Neural Networks and describe several of the recent advances in this field. The performances of recently developed Neural Network Algorithm over benchmark datasets have been tabulated. Finally, some the applications of this field have been provided.

ACM SIGCSE Bulletin, 2003

Mathematics and Computers in Simulation, 1990

ABSTRACT The ADAPT project is a collaboration of researchers in robotics, linguistics and artific... more ABSTRACT The ADAPT project is a collaboration of researchers in robotics, linguistics and artificial intelligence at three universities to create a cognitive architecture specifically designed to be embodied in a mobile robot. There are major respects in which existing cognitive architectures are inadequate for robot cognition. In particular, they lack support for true concurrency and for active perception. ADAPT addresses these deficiencies by modeling the world as a network of concurrent schemas, and modeling perception as problem solving. Schemas are represented using the RS (Robot Schemas) language, and are activated by spreading activation. RS provides a powerful language for distributed control of concurrent processes. Also, The formal semantics of RS provides the basis for the semantics of ADAPT's use of natural language. We have implemented the RS language in Soar, a mature cognitive architecture originally developed at CMU and used at a number of universities and companies. Soar's subgoaling and learning capabilities enable ADAPT to manage the complexity of its environment and to learn new schemas from experience. We describe the issues faced in developing an embodied cognitive architecture, and our implementation choices.

Procedia Manufacturing, 2015

There is a growing body of evidence that human perception is "active", in the sense that it is la... more There is a growing body of evidence that human perception is "active", in the sense that it is largely goal-oriented and top-down. Task goals appear to influence how people perceive their environment. Effective interaction between people and unmanned systems requires that the unmanned systems' perceptions be comprehensible to people, and this means that the unmanned systems should also perceive the world in an active manner. Recent evidence in cognitive psychology and neuroscience supports the proposition that simulation, the "re-enactment of perceptual, motor and introspective states" is a central cognitive mechanism. Cognitive functions such as anticipation and planning operate through a process of internal simulation of actions and environment. Indeed there is a history in the field of Artificial Intelligence of using "simulated action" as an algorithmic search procedure, e.g., game trees, though such an approach typically has problematic computational complexity. The simulations include not just the effect of actions, but also the understood laws of physics (e.g., will a falling object continue to fall). We are building a robot cognitive architecture that is based on a unified cognitive architecture-Soar-and that uses active perception and simulation in planning. Our system constructs a 3D virtual copy of itself and its environment, including people, and updates this model in realtime to agree with changes in its environment. This 3D model is a qualitative description of the world, and includes physics simulation capabilities so that effects of actions can be simulated before being executed in the real world. This allows the unmanned system to predict and plan its interactions with people. This paper describes the structure of this architecture, and provides examples and videos showing its performance in dynamic environments.

2007 IEEE Symposium on Computational Intelligence in Security and Defense Applications, 2007

The VMSoar project at Pace University is building a cognitive agent for cybersecurity. The projec... more The VMSoar project at Pace University is building a cognitive agent for cybersecurity. The project's objective is to create an intelligent agent that can model and understand the activities of users who are on the network, and that can communicate with network administrators in English to alert them to illegal or suspicious activities. VMSoar can understand users' activities because it is capable of performing these activities itself. It knows how to perform both legal and illegal activities, and uses this knowledge to explore simulations of the activity on a network. It can also probe information stored on a machine to assess the legality of past activity. Research in cybersecurity is difficult is due to the extremely large amount of data that must be analyzed to detect illegal activities. In addition, new exploits are developed frequently. Most current projects in this area are attempting to build some level of intelligence into their systems; however, those projects are focusing primarily on statistical data mining approaches. The VMSoar project is unique in its approach to building an intelligent security agent. The VMSoar agent is based on Soar, a mature cognitive architecture that is used in universities and corporations around the world. I.

We are developing a robotic cognitive architecture to be embedded in autonomous robots that can s... more We are developing a robotic cognitive architecture to be embedded in autonomous robots that can safely interact and collaborate with people on a wide range of physical tasks. Achieving true autonomy requires increasing the robot’s understanding of the dynamics of its world (physical understanding), and particularly the actions of people (cognitive understanding). Our system’s cognitive understanding arises from the Soar cognitive architecture, which constitutes the reasoning and planning component. The system’s physical understanding stems from its central representation, which is a 3D virtual world that the architecture synchronizes with the environment in real time. The virtual world provides a common representation between the robot and humans, thus improving trust between them and promoting effective collaboration.

SPIE Proceedings, 2016

We are building a cognitive vision system for mobile robots that works in a manner similar to the... more We are building a cognitive vision system for mobile robots that works in a manner similar to the human vision system, using saccadic, vergence and pursuit movements to extract information from visual input. At each fixation, the system builds a 3D model of a small region, combining information about distance, shape, texture and motion to create a local dynamic spatial model. These local 3D models are composed to create an overall 3D model of the robot and its environment. This approach turns the computer vision problem into a search problem whose goal is the acquisition of sufficient spatial understanding for the robot to succeed at its tasks. The research hypothesis of this work is that the movements of the robot's cameras are only those that are necessary to build a sufficiently accurate world model for the robot's current goals. For example, if the goal is to navigate through a room, the model needs to contain any obstacles that would be encountered, giving their approximate positions and sizes. Other information does not need to be rendered into the virtual world, so this approach trades model accuracy for speed.

Advances in Intelligent Systems and Computing, 2016

Robots interacting and collaborating with people need to comprehend and predict their movements. ... more Robots interacting and collaborating with people need to comprehend and predict their movements. We present an approach to perceiving and modeling behaviors using a 3D virtual world. The robot's visual data is registered with the virtual world to construct a model of the dynamics of the behavior and to predict future motions using a physics engine. This enables the robot to visualize alternative evolutions of the dynamics and to classify them. The goal of this work is to use this ability to interact more naturally with humans and to avoid potentially disastrous mistakes.

The effective reuse of domain theories in problem solving requires the problem-solving agent to i... more The effective reuse of domain theories in problem solving requires the problem-solving agent to identify general theories whose properties "scale up": they hold for a class of problems of varying size. Otherwise, the agent will be overwhelmed by the cost of indexing and retrieving a huge collection of domain theories, each of which applies in very restricted cases. Furthermore, these general theories need to be represented in a manner that is as independent as possible of the circumstances of particular cases. This paper describes research on analysis and reformulation of domain theories. The perspective of this work is to view a problem space as though it were physical space, and the actions in the problem space as though they were physical motions. A domain theory should then state the laws of motion within the space. Following the analogy with physics, a representation is a coordinate system, and theories are reformulated by transforming coordinates. The mathematical basis for this analogy is briefly given, and illustrated on two simple examples.

Procedia Manufacturing, 2015

There is a growing body of evidence that human perception is "active", in the sense that it is la... more There is a growing body of evidence that human perception is "active", in the sense that it is largely goal-oriented and top-down. Task goals appear to influence how people perceive their environment. Effective interaction between people and unmanned systems requires that the unmanned systems' perceptions be comprehensible to people, and this means that the unmanned systems should also perceive the world in an active manner. Recent evidence in cognitive psychology and neuroscience supports the proposition that simulation, the "re-enactment of perceptual, motor and introspective states" is a central cognitive mechanism. Cognitive functions such as anticipation and planning operate through a process of internal simulation of actions and environment. Indeed there is a history in the field of Artificial Intelligence of using "simulated action" as an algorithmic search procedure, e.g., game trees, though such an approach typically has problematic computational complexity. The simulations include not just the effect of actions, but also the understood laws of physics (e.g., will a falling object continue to fall). We are building a robot cognitive architecture that is based on a unified cognitive architecture-Soar-and that uses active perception and simulation in planning. Our system constructs a 3D virtual copy of itself and its environment, including people, and updates this model in realtime to agree with changes in its environment. This 3D model is a qualitative description of the world, and includes physics simulation capabilities so that effects of actions can be simulated before being executed in the real world. This allows the unmanned system to predict and plan its interactions with people. This paper describes the structure of this architecture, and provides examples and videos showing its performance in dynamic environments.

Multisensor, Multisource Information Fusion: Architectures, Algorithms, and Applications 2015, 2015

ABSTRACT In previous work, we have shown how a 3D model can be built in real time and synchronize... more ABSTRACT In previous work, we have shown how a 3D model can be built in real time and synchronized with the environment. This world model permits a robot to predict dyamics in its environment and classify behaviors. In this paper we evaluate the effect of such a 3D model on the accuracy and speed of various computer vision algorithms, including tracking, optical flow and stereo disparity. We report results based on the KITTI database and on our own videos.

SPIE Proceedings, 2012

We consider the scenario where an autonomous platform that is searching or traversing a building ... more We consider the scenario where an autonomous platform that is searching or traversing a building may observe unstable masonry or may need to travel over unstable rubble. A purely behaviour-based system may handle these challenges but produce behaviour that works against long-terms goals such as reaching a victim as quickly as possible. We extend our work on ADAPT, a cognitive robotics architecture that incorporates 3D simulation and image fusion, to allow the robot to predict the behaviour of physical phenomena, such as falling masonry, and take actions consonant with long-term goals. We experimentally evaluate a cognitive only and reactive only approach to traversing a building filled with various numbers of challenges and compare their performance. The reactive only approach succeeds only 38% of the time, while the cognitive only approach succeeds 100% of the time. While the cognitive only approach produces very impressive behaviour, our results indicate how much better the combination of cognitive and behaviour-based can be.

SPIE Proceedings, 2011

An important component of cognitive robotics is the ability to mentally simulate physical process... more An important component of cognitive robotics is the ability to mentally simulate physical processes and to compare the expected results with the information reported by a robot's sensors. In previous work, we have proposed an approach that integrates a 3D game-engine simulation into the robot control architecture. A key part of that architecture is the Match-Mediated Difference (MMD) operation, an approach to fusing sensory data and synthetic predictions at the image level. The MMD operation insists that simulated and predicted scenes are similar in terms of the appearance of the objects in the scene. This is an overly restrictive constraint on the simulation since parts of the predicted scene may not have been previously viewed by the robot. In this paper we propose an extended MMD operation that relaxes the constraint and allows the real and synthetic scenes to differ in some features but not in (selected) other features. Image difference operations that allow a real image and synthetic image generated from an arbitrarily colored graphical model of a scene to be compared. Scenes with the same content show a zero difference. Scenes with varying foreground objects can be controlled to compare the color, size and shape of the foreground.

Proceedings of the ACM/IEEE international conference on Human-robot interaction, 2007

The ADAPT project is a collaboration of researchers in linguistics, robotics and artificial intel... more The ADAPT project is a collaboration of researchers in linguistics, robotics and artificial intelligence at three universities. We are building a complete robotic cognitive architecture for a mobile robot designed to interact with humans in a range of environments, and which uses natural language and models human behavior. This paper concentrates on the HRI aspects of ADAPT, and especially on how ADAPT models and interacts with humans.

SPIE Proceedings, 2010

This paper describes our work on integrating distributed, concurrent control in a cognitive archi... more This paper describes our work on integrating distributed, concurrent control in a cognitive architecture, and using it to classify perceived behaviors. We are implementing the Robot Schemas (RS) language in Soar. RS is a CSP-type programming language for robotics that controls a hierarchy of concurrently executing schemas. The behavior of every RS schema is defined using port automata. This provides precision to the semantics and also a constructive means of reasoning about the behavior and meaning of schemas. Our implementation uses Soar operators to build, instantiate and connect port automata as needed. Our approach is to use comprehension through generation (similar to NLSoar) to search for ways to construct port automata that model perceived behaviors. The generality of RS permits us to model dynamic, concurrent behaviors. A virtual world (Ogre) is used to test the accuracy of these automata. Soar's chunking mechanism is used to generalize and save these automata. In this way, the robot learns to recognize new behaviors.

2008 ECSIS Symposium on Learning and Adaptive Behaviors for Robotic Systems (LAB-RS), 2008