Osiris on Chip

Embedded-systems designed for the decision making and real-time control sectors

Introduction

The teams at LSI are currently working on the modelling and representation of knowledge and data. As part of their work, the teams have developed an object-oriented system on an FPGA chip for the representation and management of data and knowledge (SGBD-BC). Named  OSIRIS on Chip, this system is a tool that is capable of reproducing the cognitive mechanisms of an expert in well under a microsecond and in a specific field.

Advantages

  • Intelligent: capable of answering questions by formulating a thought process based on known facts and rules, OSIRIS on Chip uses an inference engine that allows it to generate new facts. Thus the system gains experience over time.
  • Powerful: Combining very fast classification methods and the FPGA chip’s powerful calculation, OSIRIS on Chip enables the system to answer questions in 10 nanoseconds, i.e. a million times faster than any other software like it. Its power therefore allows real-time control in a complex environment.
  • Large capacity: As a data manager, it allows data sharing by various user categories, through different views, as well as the manipulation of large quantities of remaining data.
  • Integrated : As an intelligent embedded system, OSIRIS on Chip can be integrated into numerous mobile systems (robots, autonomous vehicles, aeroplanes). It also functions in extreme conditions.

Service Offering

  • R&D Collaboration
  • Knowledge Transfer
  • Advice

 

Technology

 

Principe de fonctionnement de Osiris on Chip

Based on a structured representation of knowledge (ontology) in the form of classes and statistical analysis of this knowledge, the OSIRIS on Chip technology organises data into a hierarchical structure of concepts that defines the restrictions on decision processing. By means of production rules, the system is able to identify and initialise objects even if it is not completely familiar with them. An original classification model inspired by neural networks enables probability classification of objects contained in the knowledge base. The information acquired therefore always generates new knowledge even if it is incomplete. The updating of knowledge in real-time allows the system to reconfigure itself dynamically.

Temps de classification Osiris on Chip

Areas of Application

  • Decision-making Assistance
  • Embedded Intelligence
  • Real-time Control
  • Artificial Intelligence

 
Contact Us