Figure 7 - uploaded by Robert Kozma
Content may be subject to copyright.
A: The dots mark EMMA's locations detected by the external camera using wallfollowing training mode in an environment with obstacles.
Source publication
We use the biologically inspired dynamic neural network architecture KIV to achieve robust goal-oriented navigation in a physical environment with obstacles. KIV operates on the principle of chaotic neurodynamics, in the style of brains. It performs the task of multi-sensory fusion, recognition, and decision-making in real time. We use the Sony AIB...
Similar publications
We show how a system for video-rate parallel camera tracking and 3D map-building can be readily extended to allow one or more cameras to work in several maps, separately or simultaneously. The ability to handle several thousand features per map at video-rate, and for the cameras to switch automatically between maps, allows spatially localized AR wo...
Citations
... Experiments indicate that the amygdala, together with the orbitofrontal cortex, is involved in decision making (Bechara, Damasio, & Damasio, * Le Doux, 2000; Zhou & Coggins, 2002). The interaction between entorhinal cortex, amygdala, hippocampal and cortical areas is studied in and Kozma and Muthu (2004). The role of entorhinal cortex in decision making under the influence of sensory, orientation, and motivational clues has been evaluated (Kozma, 2007a). ...
... Today, K sets are used in a wide range of applications, including classification (Chang, Freeman, & Burke, 1998; Freeman, Kozma, & Werbos, 2001), image recognition (Li, Lou, Wang, Li, & Freeman, 2006), time series prediction (Beliaev & Kozma, 2007), and robot navigation (Harter & Kozma, 2005; Voicu, Kozma, Wong, & Freeman, 2004). Recent developments include KIV sets for sensor fusion (Kozma & Tunstel, 2005; Kozma & Muthu, 2004), and autonomous control (Kozma, 2007a; Kozma et al., 2008). The hierarchical K model based approach is summarized in the 10 ''Building Blocks'' of neurodynamics (Freeman, 1975Freeman, , 2001): 1. Non-zero point attractor generated by a state transition of an excitatory population starting from a point attractor with zero activity. ...
Human cognition performs granulation of the seemingly homogeneous temporal sequences of perceptual experiences into meaningful and comprehensible chunks of fuzzy concepts and behaviors. These knowledge granules are stored and consequently accessed during action selection and decisions. A dynamical approach is presented here to interpret experimental findings using K (Katchalsky) models. In the K model, meaningful knowledge is repetitiously created and processed in the form of sequences of oscillatory patterns of neural activity distributed across space and time. These patterns are not rigid but flexible and intermittent; soon after they arise through phase transitions, they dissipate. Computational implementations demonstrate the operation of the model based on the principles of intentional brain dynamics.
... Further successful demonstrations of the KIV-based control are given using a simulated 2D Martian environment [38], as well as using the Sony Aibo ERS 220 mobile robot platform [39]. Preliminary results using NASA SRR2K rover have been reported in [28,40,41]. ...
... Here a brief description of the system is given. For details of the SODAS operation, see [32], [35], [39]. the stored reference activations. ...
Intentional behavior is a basic property of intelligence, and it incorporates the cyclic operation of prediction, testing by action, sensing, perceiving and assimilating the experienced features. Intentional neurodynamic principles are applied for on-line processing of multisensory inputs and for the generation of dynamic behavior using the SRR (Sample Return Rover) platform at the indoor facility of the Planetary Robotics Laboratory, Jet Propulsion Laboratory. The studied sensory modalities include CMOS camera vision, orientation based on an internal motion unit and accelerometer signals. The control architecture employs a biologically inspired dynamic neural network operating on the principle of chaotic neural dynamics manifesting intentionality in the style of mammalian brains. Learning is based on Hebbian rules coupled with reinforcement. The central issue of this work is to study how the developed control system builds associations between the sensory modalities to achieve robust autonomous action selection. The proposed system builds such associations in a self-organized way and it is called Self-Organized Development of Autonomous Adaptive Systems (SODAS). This system operates autonomously, without the need for human intervention, which is a potentially very beneficial feature in challenging environments, such as encountered in space explorations at remote planetary environments. The experiments illustrate obstacle avoidance combined with goal-oriented navigation by the SRR robot using SODAS control principles.
... In a given task, a memory could be 3-4 steps deep, or more. One can optimize the system performance based on the memory depth [109]. Such optimization has not been done in the SRR2K experiments; it is planned to be completed in the future. ...
We present an intentional neurodynamic theory for higher cognition and intelligence. This theory provides a unifying framework for integrating symbolic and subsymbolic methods as complementary aspects of human intelligence. Top-down symbolic approaches benefit from the vast experience with logical reasoning and with high-level knowledge processing in humans. Connectionist methods use bottom-up approach to generate intelligent behavior by mimicking subsymbolic aspects of the operation of brains and nervous systems. Neurophysiological correlates of intentionality and cognition include sequences of oscillatory patterns of mesoscopic neural activity. Oscillatory patterns are viewed as intermittent representations of generalized symbol systems, with which brains compute. These dynamical symbols are not rigid but flexible. They disappear soon after they emerged through spatio-temporal phase transitions. Intentional neurodynamics provides a solution to the notoriously difficult symbol grounding problem. Some examples of implementations of the corresponding dynamic principles are described in this review.
... Further successful demonstrations of the KIV-based control are given using a simulated 2D Martian environment [38], as well as using the Sony Aibo ERS 220 mobile robot platform [39]. Preliminary results using NASA SRR2K rover have been reported in [28,40,41]. ...
... Here a brief description of the system is given. For details of the SODAS operation, see [32], [35], [39]. the stored reference activations. ...
... The results compare very well with Vershure's results in the original Distributed Adaptive Control experiment [36], and with the object avoidance performance of Schmitt trigger [37]. Further successful demonstrations of the KIV-based control are given using a simulated 2D Martian environment [38], as well as using the Sony Aibo ERS 220 mobile robot platform [39]. ...
... Here a brief description of the system is given. For details of the SODAS operation, see [39]. 1. At the learning phase, SRR explores the environment and builds association between visual, IMU, and orientation sensory modalities. ...
We demonstrate the operation of the SODAS approach (self-organized ontogenetic development of autonomous systems) for on-line processing of sensory inputs and onboard dynamic behavior tasking using SRR2K (sample return rover) platform at the planetary robotics indoor facility of JPL. SODAS employs a biologically inspired dynamic neural network architecture operating on the principle of chaotic neural dynamics manifesting intentionality in the style of brains. Intentional behavior includes the cyclic operation of prediction, testing by action, sensing, perceiving, and assimilating the learned features. The experiments illustrate robust obstacle avoidance combined with goal-oriented navigation by the SRR2K robot.
... If proper learning and generalization took place, one could change the terrain into a layout which SRR2K had never seen before; still should achieve good performance. The KIV-guided robot uses its experience to continuously solve problems in perception and navigation that are imposed by its environment as it pursues autonomously the goals selected by its trainer (Kozma, Muthu, 2004). ...
... In the given task, a memory could be 3-4 steps deep, or more. One can in fact optimize the learning and system performance based on the memory depth (Kozma & Muthu, 2004). This has not been used in the SRR2K experiments yet, but is planned to be completed in the future. ...
We review computational intelligence methods of sensory perception and cognitive functions in animals, humans, and artificial devices. Top-down symbolic methods and bottom-up sub-symbolic approaches are described. In recent years, computational intelligence, cognitive science and neuroscience have achieved a level of maturity that allows integration of top-down and bottom-up approaches in modeling the brain. Continuous adaptation and teaming is a key component of computationally intelligent devices, which is achieved using dynamic models of cognition and consciousness. Human cognition performs a granulation of the seemingly homogeneous temporal sequences of perceptual experiences into meaningful and comprehensible chunks of concepts and complex behavioral sehemas. They are accessed during action selection and conscious decision making as part of the intentional cognitive cycle. Implementations in computational and robotic environments are demonstrated.
... There are three phases in the operation of KIV control: (i) learning phase; (ii) labeling phase; and (iii) testing phase. For details see (Kozma, Muthu, 2004). ...
... In the given task, a memory could be 3-4 steps deep, or more. One can in fact optimize the learning and system performance based on the memory depth (Kozma & Muthu, 2004). This has not been used in the SRR2K experiments yet and planned to be completed in the future. ...
The chapter reviews mechanisms of generation and utilization of knowledge in human cognitive activity and in artificial intelligence systems. First we explore experience-based methods, including top-down symbolic approaches, which address knowledge processing in humans. Symbolic theories of intelligence fall short of explaining and implementing strategies routinely produced by human intelligence. Connectionist methods are rooted in our understanding of the operation of brains and nervous systems, and they gain popularity in constructing intelligent devices. Contrary to top-down symbolic methods, connectionism uses bottom-up emergence to generate intelligent behaviors. Recently, computational intelligence, cognitive science and neuroscience have achieved a level of maturity that allows integration of top-down and bottom-up approaches, in modeling the brain.
... If proper learning and generalization took place, one could change the terrain into a layout which SRR2K had never seen before; still should achieve good performance. The KIV-guided robot uses its experience to continuously solve problems in perception and navigation that are imposed by its environment as it pursues autonomously the goals selected by its trainer (Kozma, Muthu, 2004). ...
... In the given task, a memory could be 3-4 steps deep, or more. One can in fact optimize the learning and system performance based on the memory depth (Kozma & Muthu, 2004). This has not been used in the SRR2K experiments yet, but is planned to be completed in the future. ...
The conditions and mechanisms for producing general intelligent action in agents are the focus and proper study of cognitive science. Many such positions have been proposed, including symbolic and connectionist viewpoints. New points of view are beginning to emerge, including embodied and dynamical cognition, but have not yet been fully solidified into a single comprehensive position. In this article we present one such new viewpoint that emphasizes the importance of nonconvergent dynamics to the production of general intelligent behavior. This approach represents a fourth generation of connectionist thought, and is informed from new results in neuroscience and computational neurodynamics. We formulate the necessary and sufficient conditions for the production of intelligent behavior in this approach to cognition and introduce one such model capable of meeting these conditions. Keywords computational neuroscience, cognitive architecture, nonlinear dynamical systems Nonconvergent Dynamics and Cognitive Systems What are the mechanisms by which biological organisms, including human beings, produce general intelligent actions in order to survive, reproduce and thrive in their environment? This, in some form, is one of the basic questions that lies at the heart of cognitive science. Various ideas have been put forward as possible answers to this question. Metaphors of cognition have been inspired from the advanced technologies of the times, including hydraulic, phone switchboard and computer metaphors (Von Neumann, 1958). Inspiration has been sought from the realms of formal logic as the means of general intelligent action. Others have looked to abstracted models of how neural tissue functions as possibly holding the key insights to the production of intelligent behavior. Hi...
