Defence Artificial Intelligence Research Network (DAIRNet) Research Call
The Next Generation Technologies Fund (NGTF) seeks proposals from academia, industry and other government research agencies in research activities across a range of AI-related focus areas.
This program of work aims to use artificial intelligence to process noisy and dynamic data in order to produce outcomes to provide decision superiority to Australian Defence. This may include the ability to fuse data and construct causal models from a wide range of modalities and diverse forms, in large volumes, and collected at varied rates. The diverse data types may range from physically measured sensor data to discrete or tokenised data such as in natural language documents.
The focus of this call is Patterns in Noisy and Dynamic Data.
Submissions are now closed.
Defence Artificial Intelligence Research Network (DAIRNet) Call
This specific call is on the problem of finding patterns in noisy and dynamic data using Artificial Intelligence.
Information (Data) patterns may be unclear and may indicate anomalies that need to be investigated or events of interest such as indications of a pandemic or an attack.
The data often will be from multiple sources, in diverse forms, in large volumes and collected at varied rates and may need to be combined before the patterns are manifest.
Proposals for up to 2.5-year research plans will be considered and successful proposals may be funded up to $100,000 in the first six months and possibly up to $500,000 in subsequent years.
Experimental findings and prototyped concepts are the expected outcomes.
What are the research areas?
Several use cases have been gathered by the DSTG research teams, and provide examples for this general research area.
- The detection of the presence of chemical or biological threats and the prediction of how they are spreading may not be apparent until effects are observed in humans. Observations of subtle variations in biomarkers may lead to inferences that the human body has been exposed to toxic chemicals or dangerous biological pathogens. An ambitious aim is to detect such events before poisoning or infection has harmful effects, i.e. very early after exposure, to maximise the effectiveness of possible interventions. Major challenges arise from the fact that there are many causes for subtle biomarker variations; that these variations are not random; and that different humans react differently to threat exposure. Thus, it is important to develop new accurate and differentiating data analytics and inferencing methods in parallel with novel wearable sensor technologies;
- Significant amounts of data are collected for each flight an aircraft undergoes. In addition to the volume of data, there is a variety of data related to flight dynamics and various on-board sensors. One challenge is to use machine learning across this disparate data, looking for the signals that indicate anomalous events such as structural problems and predict times to failure in order to maximise uptime and availability of assets;
- Missions involving multiple agent coordination (involving for example unmanned aerial vehicles and unmanned ground vehicles) in complex environments will utilise an array of different sensing modalities and information sources including text, imagery from radar, infra-red and lidar. The size limitations of these agents requires edge computing – algorithms that are relatively fast, use small memory and low power. An exemplar mission is finding hidden adversaries in congested, dynamic settings (e.g. urban) where information is multi-modal and requires fusion; objects of interest are moving and need to be tracked within limited fields of view; with operations at close range necessitating decisions being made in short timeframes. Object detection and tracking may necessitate moving sensors based on 3-D representations of the space of interest. Compute limitations may necessitate processing on other platforms. One aspiration is to automatically describe these complex scenes using visual narratives.
Submissions are to be made via AusTender.
DSTG will provide a research program brief to all interested parties via Microsoft Teams on 14 October 2021 at 2pm Australian Eastern Daylight Time (AEDT). This will provide an opportunity for clarifications and questions to be raised. Any further questions after this brief will need to be provided in writing with DSTG providing a written response that will be provided to all parties. Nominations to attend the brie1fing are to be forwarded in writing to the IISR mailbox: VVFE@qfg.qrsrapr.tbi.nh to be provided with a Microsoft Teams invite.
For security purposes: Full Name, Organisation, Position, Mobile Telephone No. and email address of all nominated personnel are to be included at the time of nomination
Contact Officer and RFP Inquiries
Proposers are to direct any questions or concerns regarding this RFP in writing to the Contact Officer.
Proposers may submit questions or concerns to the Contact Officer up until five Working Days prior to the Closing Time specified in the Tender Details Schedule.
The call for proposals closes at 1400 (AEDT) on 11 November 2021.
Tender responses are to be submitted electronically via the AusTender website using the ‘Lodgement Page’ link on the ‘tender notification’.
AusTender provides an online automatic notification facility for registered users who have registered particular areas of interest in product and service categories. Registered users will then be notified of new business opportunities and planned procurements via email once business opportunities that match their specified profile are published. A registered user can edit their profile at any time to amend their notification profile or disable automatic email notification if it is no longer required.