Objective
LOCUST aims to both achieve substantial scientific and technological progress in several complementary knowledge areas required for the realization of bio-inspired Vision Systems on Chip (VSoCs), and to develop generally valid VSoC design techniques. The resulting biologically inspired architectures rely exclusively on visual information, and their output may be used by any autonomously acting agent. Moreover, a corresponding single-chip implementation can be expected to have a high impact on a wide field of applications, due to significant improvements in performance and low-cost hardware. Specifically, the consortium will focus on the visual system of locusts, where looming-sensitive neurons were identified whose activity correlate with corresponding escape and avoidance behaviour, respectively. The idea is to unravel and model the underlying neural circuitry in order to realize a prototype of a mixed-signal VLSI VSoC chip featuring visual sensors, parallel processing, and decision making on one single chip. This VSoC will be the core of a collision-threat warning module for automotive application.
OBJECTIVES
The main objective is to gain knowledge and to progress towards the practical exploitation of artificial Vision Systems on Chips (VSoC) based on the emulation of natural vision systems. As a consequence, the consortium is remarkably multidisciplinary, comprising neurobiologists, mixed signal VLSI designers, information-technology experts, and industrial partners. In addition to the development of general guidelines and design techniques for arbitrary VSoCs, the project aims to produce a demonstrator for collision-threat warning in automobiles, capable of operating under the whole range of conditions typical of automotive applications, and complying with the strict reliability rules of this industry. For this purpose, the consortium will focus on the emulation of the locusta migratoria biological circuitry for looming alert. The developed system will be installed in an automobile, and will be tested under real-life conditions.
DESCRIPTION OF WORK
The work is structured into 6 Work Packages. WP1 to WP4 embrace the technical issues, while WP0 is dedicated to management and quality assurance, and WP5 to dissemination and exploitation plans. The work packages are the following:
WP0: Project Management and Quality Assurance;
WP1: Biological Modelling and Model Customisation;
WP2: System Architecture and Hardware/Software Integration;
WP3: Bio-Inspired Sensory-Processing Application-Specific Integrated Circuits;
WP4: System requirements and Demonstrator;
WP5: Dissemination and Exploitation.
WP1 is devoted to the modelling of the biological circuitry, using both physiological and computational methods to extract unique features of the locust visual system. Similar models from other insects will be analysed and their general characteristics will be explored. This task is led by experts in neurobiology.
WP2, led by an optoelectronics information-technology expert embraces work in the fields of modelling and implementation from a systems perspective. Activities here are aimed to determine an optimum architecture and partitioning of artificial vision systems on a chip.
WP3, led by mixed-signal IC design experts, is devoted to define suitable Vision System on Chip (VSoC) architectures, to explore the different trade-offs which are associated with their implementation, to address all pertinent robustness issues, and finally to design a VSoC prototype for the application of collision-threat warning in automobiles.
WP4, led by an automobile manufacturing company, is devoted to the definition of the specific system requirements, to embed them in a car, and evaluating the resulting real-life demonstrator. In addition to the demonstrator, the consortium will dedicate substantial efforts to the creation of generic design techniques, which are valid for any similar application-specific VSoC.
Fields of science (EuroSciVoc)
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
- natural sciencesbiological sciencesneurobiology
- natural sciencesphysical scienceselectromagnetism and electronicsoptoelectronics
- natural sciencescomputer and information sciencesartificial intelligencecomputer visionobject detection
- natural sciencescomputer and information sciencescomputational science
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensors
You need to log in or register to use this function
Topic(s)
Call for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
28006 MADRID
Spain