The challenge

Comprehending a situation from a current set of observations

By 'situation analysis' we mean the comprehension of a current situation from a given set of observations and assessing it's consequences. Comprehensive situation analysis is indispensable when making good decisions, but building software to deliver situation analysis is difficult for a variety of reasons.  In the real world we rarely have complete and/or correct observations. Sensor input like that from GPS data is inherently noisy, object recognition from video/photos is unreliable, databases are incomplete or inconsistent, and status reports from human actors can be entirely missing or be late etc.

Existing enterprise automation systems lack the capability for deeper inferences, which is needed for properly achieving situation analysis. One should expect the system to be able to fill in unobserved events, auto-correct erroneous data, retro-fit late observations to explain and to re-adjust earlier conclusions and much more.

Our response

Exploiting and advancing research using AI

The long-term goal of our approach is to achieve situation analysis capabilities by exploiting and advancing research from relevant AI areas such as automated reasoning, diagnosis, temporal logic and problemistic reasoning and making them available through a re-usable modelling language as part of the Fusemate system.

The results

Implementing a decision support system

Our project delivers an implemented decision support system and method for achieving situation analysis for a wide range of application domains. When deployed, we expect it will help to avoid delays (e.g factory floor), to support trust and detect fraud (e.g food supply chain) and generally to make better informed decisions (e.g disaster recovery).

Fusemate software development and application diagram showing three distinct sections that include System Architecture on one side leading through to Fusemate which in turn leads back and forth to Q/A/C (explained below).

System Architecture includes:

  • central column of What? When? Where? Why? with a variety of variables flowing around the centre (real world goods) and feeding into the centre column (EPC/barcode data):
    • supplier
      • raw materials
    • warehouse
      • raw materials
    • manufacturer
      • finished goods
    • distribution centre
      • finished goods
    • retailer

Fusemate contains:

  • Explanations; Interference Engine
  • Domain Model; If-then rules
  • Update
  • Rest

Domain Model - e.g., If item I is unpacked from container C at a time then I was packed into C at some time S<T.

EPCIS Events - EPCIS events are sent to fusemate as they become available.

Explanations - The inference engine derives a set of plausible models consistent with the EPCIS events so far.

Q/A - Q: Where was item I at time? A: Item I was unpacked from container C at time T and loc L.

Update - Plausible models are unpacked on every new EPCIS event ad command provided by user.

We are seeking to develop a full methodology for situation analysis in industrial operations. This goes beyond an industrial-strength implementation and includes a systematic approach to understanding a concrete domain, instantiating our system for the use case at hand, and making it available for use by non-experts. Fusemate has also been extended and is currently developed within a SIEF-funded food supply chain project.

Do business with us to help your organisation thrive

We partner with small and large companies, government and industry in Australia and around the world.

Contact us now to start doing business

Contact Data61

Your contact details

First name must be filled in

We'll need to know what you want to contact us about so we can give you an answer.