CAVIAR: A 45k Neuron, 5M Synapse, 12G Connects/s AER Hardware Sensory–Processing– Learning–Actuating System for High-Speed Visual Object Recognition and Tracking

@article{SerranoGotarredona2009CAVIARA4,
  title={CAVIAR: A 45k Neuron, 5M Synapse, 12G Connects/s AER Hardware Sensory–Processing– Learning–Actuating System for High-Speed Visual Object Recognition and Tracking},
  author={Rafael Serrano-Gotarredona and Matthias Oster and Patrick Lichtsteiner and Alejandro Linares-Barranco and Rafael Paz-Vicente and Francisco Gomez-Rodriguez and Luis A. Camu{\~n}as-Mesa and Raphael Berner and Manuel Rivas P{\'e}rez and Tobi Delbr{\"u}ck and Shih-Chii Liu and Rodney J. Douglas and Philipp H{\"a}fliger and Gabriel Jim{\'e}nez-Moreno and Antonio Abad Civit Balcells and Teresa Serrano-Gotarredona and Antonio Acosta-Jim{\'e}nez and Bernab{\'e} Linares-Barranco},
  journal={IEEE Transactions on Neural Networks},
  year={2009},
  volume={20},
  pages={1417-1438}
}
This paper describes CAVIAR, a massively parallel hardware implementation of a spike-based sensing-processing-learning-actuating system inspired by the physiology of the nervous system. CAVIAR uses the asynchronous address-event representation (AER) communication framework and was developed in the context of a European Union funded project. It has four custom mixed-signal AER chips, five custom digital AER interface components, 45 k neurons (spiking cells), up to 5 M synapses, performs 12 G… 
A Real-Time, Event-Driven Neuromorphic System for Goal-Directed Attentional Selection
TLDR
A flexible event-driven platform consisting of the integration of a visual AER sensor and the SpiNNaker system, a programmable massively parallel digital architecture oriented to the simulation of spiking neural networks, and the implementation of a neural network for feature-based attentional selection on this platform are presented.
Neuro-inspired system for real-time vision sensor tilt correction
TLDR
An AER based layer able to correct in real time the tilt of an AER vision sensor, using a high speed algorithmic mapping layer is presented.
Neocortical frame-free vision sensing and processing through scalable Spiking ConvNet hardware
This paper summarizes how Convolutional Neural Networks (ConvNets) can be implemented in hardware using Spiking neural network Address-Event-Representation (AER) technology, for sophisticated pattern
On scalable spiking convnet hardware for cortex-like visual sensory processing systems
This paper summarizes how Convolutional Neural Networks (ConvNets) can be implemented in hardware using Spiking neural network Address-Event-Representation (AER) technology, for sophisticated pattern
Spike-Based Convolutional Network for Real-Time Processing
TLDR
The first bio-inspired six layer convolutional network (ConvNet) non-frame based that can be implemented with already physically available spikebased electronic devices is proposed and simulation results show recognition delays of 16 milliseconds from stimulus onset.
AER spike-processing filter simulator: Implementation of an AER simulator based on cellular automata
TLDR
A software simulator able to gather several spike-based elements into the same workspace in order to test a CA architecture based on AER before a hardware implementation is presented.
An AER Spike-Processing Filter Simulator and Automatic VHDL Generator Based on Cellular Automata
TLDR
A software simulator able to gather several spike-based elements into the same workspace in order to test a CA architecture based on AER before a hardware implementation, and produces VHDL for testing the AER-CA into the FPGA of the USBAER A ER-tool.
An AER to CAN Bridge for Spike-Based Robot Control
TLDR
This paper presents a co-design FPGA and embedded computer system that implements a bridge between these two protocols: CAN and AER.
The ripple pond: enabling spiking networks to see
TLDR
Key aspects in the proposed approach include utilizing the spatial properties of physically embedded neural networks and propagating waves of activity therein for information processing, using dimensional collapse of imagery information into amenable TP and the use of asynchronous frames for information binding.
High-speed serial interfaces for event-driven neuromorphic systems
TLDR
On-going progress is shown on bit-serial SATA AER inter-FPGA communications for multi-tile scalable neuromorphic systems and address-Event-Representation, a technology widely used among neuromorphic engineers to emulate such massive interconnectivity among neurons by time-multiplexing fast digital channels.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 113 REFERENCES
A Multichip Pulse-Based Neuromorphic Infrastructure and Its Application to a Model of Orientation Selectivity
TLDR
A general-purpose solution in the form of a peripheral component interconnect (PCI) board (the PCI-AER board) supported by software is implemented and a reconfigurable multichip neuromorphic system for feature selectivity which models orientation tuning properties of cortical neurons is demonstrated.
A Neuromorphic Cortical-Layer Microchip for Spike-Based Event Processing Vision Systems
TLDR
A neuromorphic cortical-layer processing microchip for address event representation (AER) spike-based processing systems that computes convolutions of programmable kernels over the AER visual input information flow and allows for a bio-inspired coincidence detection processing.
High-speed character recognition system based on a complex hierarchical AER architecture
TLDR
The paper describes briefly cortex-like spiking vision processing principles, and the AER (address event representation) technique used in hardware spiking systems, and an example application is described, which is a simplification of Fukushima's Neocognitron.
An event-based VLSI network of integrate-and-fire neurons
TLDR
A mixed-mode VLSI device that comprises a reconfigurable network of I&F neurons and adaptive synapses, and it is shown that a competitive network configuration has winner-take-all (WTA) behaviour and produces spike synchronization.
Adaptive WTA With an Analog VLSI Neuromorphic Learning Chip
  • P. Häfliger
  • Computer Science
    IEEE Transactions on Neural Networks
  • 2007
TLDR
It is demonstrated how a particular spike-based learning rule can be tuned to express rate-based classical Hebbian learning behavior, where the average input and output spike rates are sufficient to describe the synaptic changes.
On Real-Time AER 2-D Convolutions Hardware for Neuromorphic Spike-Based Cortical Processing
TLDR
A chip that performs real-time image convolutions with programmable kernels of arbitrary shape is presented, and discussions and results on scaling up the approach for larger pixel arrays and multilayer cortical AER systems are provided.
Address-event-based platform for bioinspired spiking systems
TLDR
This paper addresses the problem of communicating several AER based chips to compose a powerful processing system, based basically on an embedded computer, a powerful FPGA and serial links, to make the system faster and be stand alone (independent from a PC).
Fully digital AER convolution chip for vision processing
TLDR
A neuromorphic fully digital convolution microchip for address event representation (AER) spike-based processing systems that computes 2-D convolutions with a programmable kernel in real time.
Quantification of a Spike-Based Winner-Take-All VLSI Network
TLDR
A theoretical description of the WTA computation which takes into consideration the input statistics, neuron response variance, and output rates of spike-based winner-take-all VLSI chips is shown.
...
1
2
3
4
5
...