Perception-Action Cycle Models, Architectures, and Hardware
:
John G. Taylor, Amir Hussain, Vassilis Cutsuridis
:
Vassilis Cutsuridis, Amir Hussain, John G. Taylor
:
Perception-Action Cycle Models, Architectures, and Hardware
:
Springer-Verlag
:
9781441914521
:
1
:
CHF 301.60
:
:
Nichtklinische Fächer
:
English
:
784
:
Wasserzeichen
:
PC/MAC/eReader/Tablet
:
PDF
The perception-action cycle is the circular flow of information that takes place between the organism and its environment in the course of a sensory-guided sequence of behaviour towards a goal. Each action causes changes in the environment that are analyzed bottom-up through the perceptual hierarchy and lead to the processing of further action, top-down through the executive hierarchy, toward motor effectors. These actions cause new changes that are analyzed and lead to new action, and so the cycle continues. The Perception-action cycle: Models, architectures and hardware book provides focused and easily accessible reviews of various aspects of the perception-action cycle. It is an unparalleled resource of information that will be an invaluable companion to anyone in constructing and developing models, algorithms and hardware implementations of autonomous machines empowered with cognitive capabilities. The book is divided into three main parts. In the first part, leading computational neuroscientists present brain-inspired models of perception, attention, cognitive control, decision making, conflict resolution and monitoring, knowledge representation and reasoning, learning and memory, planning and action, and consciousness grounded on experimental data. In the second part, architectures, algorithms, and systems with cognitive capabilities and minimal guidance from the brain, are discussed. These architectures, algorithms, and systems are inspired from the areas of cognitive science, computer vision, robotics, information theory, machine learning, computer agents and artificial intelligence. In the third part, the analysis, design and implementation of hardware systems with robust cognitive abilities from the areas of mechatronics, sensing technology, sensor fusion, smart sensor networks, control rules, controllability, stability, model/knowledge representation, and reasoning are discussed.
726
264
22.6.1 Continuous Restricted Boltzmann Machine
726
22.6.1.1 Continuous Stochastic Neuron
726
22.6.1.2 CRBM Learning Rule
727
22.6.2 Training Methodology
728
22.6.3 Simulation Results
729
22.6.3.1 With Simple, Multidimensional Overlapping Clusters
730
22.6.3.2 With 2D Non-Gaussian Meshed Clusters
732
22.6.3.3 With Real Drifting Data
734
22.7 CRBM Hardware and Experimental Results
737
22.7.1 Chip Implementation
737
22.7.2 Learning in Hardware
738
22.7.3 Regenerating Data With a Symmetric Distribution
740
22.7.4 Regenerating Data with a Nonsymmetric Distribution
741
22.7.5 Regenerating Data with a Doughnut-Shaped Distribution
742
22.8 Discussion and Future Works
743
22.9 Summary
745
References
745
23 Bio-Inspired Mechatronics and Control Interfaces
750
23.1 Overview
750
23.2 Previous Work
752
23.3 System Architecture
754
23.3.1 Background and Problem Definition
754
23.3.2 System Training Phase
754
23.3.2.1 Recording Arm Motion
755
23.3.2.2 Recording Muscle Activity
756
23.3.3 Data Representation
757
23.3.4 Decoding Arm Motion from EMG Signals
759
23.3.5 Modeling Human Arm Movement
761
23.3.5.1 Graphical Models
761
23.3.5.2 Building the Model
762
23.3.5.3 Inference Using the Graphical Model
765
23.3.6 Filtering Motion Estimates Using the Graphical Model
766
23.3.7 Robot Control
766
23.4 Experimental Results
768
23.4.1 Hardware and Experiment Design
768
23.4.2 Efficiency Assessment
769
23.5 Conclusion and Future Extensions
772
References
774
Index
777
