Artificial Intelligence/Machine Learning, ASA(ALT), Direct to Phase II | Phase I

Artificial Intelligence/ Machine Learning (AI/ML) Focused Open Topic

Release Date: 06/11/2024
Solicitation: 24.4
Open Date: 08/12/2024
Topic Number: A244-P037
Application Due Date: 09/17/2024
Duration: 6-24 months
Close Date: 09/17/2024
Amount Up To: $250,000-$2 million

Objective

The purpose of the AI/ML Focused Open Topic is to bring potentially valuable small business innovations to the Army and create an opportunity to expand the relevance of the Army Small Business Innovation Research program to firms who do not normally compete for SBIR awards.

Description

This open topic accepts both Phase I and Direct to Phase II submissions. Phase I proposals are accepted for a cost up to $250,000 for a 6-month period of performance and Direct to Phase II proposals are accepted for a cost up to $2,000,000 for a 24-month period of performance. All submissions must address the following 6 AI sub-fields:

  • Synthetic data generation in a format applicable to a given situation that is not obtained by direct measurement. This includes visual, textual, video, geospatial, and sensor data.
  • Data Validation and Verification: Develop novel techniques for data validation and verification in a contested space where the adversary can tamper with, deny, or otherwise manipulate collected data that will ultimately be used for training or fine-tuning machine learning models. This functionality would serve to predict the next attack for better future prevention.
  • Methodologies to identify and mitigate AI risk (operational and supply chain) by quantifying and adjusting the level or human vs. automation in model development, training, testing, and deployment phases. Including authentication techniques as a form of model provenance and access control.
  • Develop new ways of implementing, constructing, and testing Large Language Models (LLM) or Radio Frequency (RF) signal detection models, their prompts, and system design that make use of these models in less time by standardizing Application Programming Interfaces (API), evaluation pipelines, prompt discovery and tuning and implementing diverse performance constraints.
  • Retrieval augmented generation (RAG) proof of concept techniques and early prototypes to enhance the accuracy and reliability of generative AI models. Specific areas of focus can include techniques for model optimization and reducing compute resources, methods to mitigate model bias with RAG, and scalable techniques for adoption of RAG.
  • Collaborative AI technologies or algorithms that enable communication between autonomous and/or semi-autonomous systems at extended ranges. Specific focus areas could include terrain shaping obstacles, ML algorithms to adapt to changing environments throughout a mission, and multi-node communication and system integration technologies.

Phase I

Phase I Submission Materials 

  • 5-page technical volume for down-select.
  • 8-slide commercialization plan; template provided in announcement.
  • “Statement of Work” outlining intermediate and final anticipated deliverables during the Phase I award period.

Post-Phase I Deliverables:

  • Small Business: A feasibility study to demonstrate the technical and commercial practicality of the concept to include an assessment of its technical readiness and potential applicability to military and commercial markets.

Phase II

Direct to Phase II Submission Materials 

  • 10-page technical volume for down-select to include a maximum of 2 pages showing how technical feasibility has already been achieved.
  • 8-slide commercialization plan; template provided in announcement.
  • “Statement of Work” outlining intermediate and final anticipated deliverables during the Phase II award period. During Phase II, firms must produce prototype solutions that will be practical and feasible to operate in edge and austere environments. Companies will provide a technology transition and commercialization plan for DOD and commercial markets. The Army will evaluate each product in a realistic field environment and provide solutions to stakeholders for further evaluation. Based on Soldier field evaluations, companies will be requested to update the previously delivered prototypes to meet final design configuration.

Phase III

Complete the maturation of the company’s technology developed in Phase II to TRL 6/7 and produce prototype to support further development and commercialization. The Army will evaluate each product in a realistic field environment and provide small solutions to stakeholders for further evaluation. Based on soldier evaluations in the field, companies will be requested to update the previously delivered prototypes to meet final design configuration.

Submission Information

For more information, and to submit your full proposal package, visit the DSIP Portal.

SBIR|STTR Help Desk: usarmy.sbirsttr@army.mil

A244-P037 Phase I and Direct to Phase II

References:

  • An Analysis of RF Transfer Learning Behavior Using Synthetic Data.
  • Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.
  • Improving the Domain Adaptation of Retrieval Augmented Generation (RAG) Models for Open Domain
  • Question Answering.
  • Human–Computer Interaction Cognitive Behavior Modeling of Command-and-Control Systems.
  • Risk-based data validation in machine learning-based software systems.
  • Signal Detection and Classification in Shared Spectrum: A Deep Learning Approach.

Objective

The purpose of the AI/ML Focused Open Topic is to bring potentially valuable small business innovations to the Army and create an opportunity to expand the relevance of the Army Small Business Innovation Research program to firms who do not normally compete for SBIR awards.

Description

This open topic accepts both Phase I and Direct to Phase II submissions. Phase I proposals are accepted for a cost up to $250,000 for a 6-month period of performance and Direct to Phase II proposals are accepted for a cost up to $2,000,000 for a 24-month period of performance. All submissions must address the following 6 AI sub-fields:

  • Synthetic data generation in a format applicable to a given situation that is not obtained by direct measurement. This includes visual, textual, video, geospatial, and sensor data.
  • Data Validation and Verification: Develop novel techniques for data validation and verification in a contested space where the adversary can tamper with, deny, or otherwise manipulate collected data that will ultimately be used for training or fine-tuning machine learning models. This functionality would serve to predict the next attack for better future prevention.
  • Methodologies to identify and mitigate AI risk (operational and supply chain) by quantifying and adjusting the level or human vs. automation in model development, training, testing, and deployment phases. Including authentication techniques as a form of model provenance and access control.
  • Develop new ways of implementing, constructing, and testing Large Language Models (LLM) or Radio Frequency (RF) signal detection models, their prompts, and system design that make use of these models in less time by standardizing Application Programming Interfaces (API), evaluation pipelines, prompt discovery and tuning and implementing diverse performance constraints.
  • Retrieval augmented generation (RAG) proof of concept techniques and early prototypes to enhance the accuracy and reliability of generative AI models. Specific areas of focus can include techniques for model optimization and reducing compute resources, methods to mitigate model bias with RAG, and scalable techniques for adoption of RAG.
  • Collaborative AI technologies or algorithms that enable communication between autonomous and/or semi-autonomous systems at extended ranges. Specific focus areas could include terrain shaping obstacles, ML algorithms to adapt to changing environments throughout a mission, and multi-node communication and system integration technologies.

Phase I

Phase I Submission Materials 

  • 5-page technical volume for down-select.
  • 8-slide commercialization plan; template provided in announcement.
  • “Statement of Work” outlining intermediate and final anticipated deliverables during the Phase I award period.

Post-Phase I Deliverables:

  • Small Business: A feasibility study to demonstrate the technical and commercial practicality of the concept to include an assessment of its technical readiness and potential applicability to military and commercial markets.

Phase II

Direct to Phase II Submission Materials 

  • 10-page technical volume for down-select to include a maximum of 2 pages showing how technical feasibility has already been achieved.
  • 8-slide commercialization plan; template provided in announcement.
  • “Statement of Work” outlining intermediate and final anticipated deliverables during the Phase II award period. During Phase II, firms must produce prototype solutions that will be practical and feasible to operate in edge and austere environments. Companies will provide a technology transition and commercialization plan for DOD and commercial markets. The Army will evaluate each product in a realistic field environment and provide solutions to stakeholders for further evaluation. Based on Soldier field evaluations, companies will be requested to update the previously delivered prototypes to meet final design configuration.

Phase III

Complete the maturation of the company’s technology developed in Phase II to TRL 6/7 and produce prototype to support further development and commercialization. The Army will evaluate each product in a realistic field environment and provide small solutions to stakeholders for further evaluation. Based on soldier evaluations in the field, companies will be requested to update the previously delivered prototypes to meet final design configuration.

Submission Information

For more information, and to submit your full proposal package, visit the DSIP Portal.

SBIR|STTR Help Desk: usarmy.sbirsttr@army.mil

References:

  • An Analysis of RF Transfer Learning Behavior Using Synthetic Data.
  • Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.
  • Improving the Domain Adaptation of Retrieval Augmented Generation (RAG) Models for Open Domain
  • Question Answering.
  • Human–Computer Interaction Cognitive Behavior Modeling of Command-and-Control Systems.
  • Risk-based data validation in machine learning-based software systems.
  • Signal Detection and Classification in Shared Spectrum: A Deep Learning Approach.

A244-P037 Phase I and Direct to Phase II

Artificial Intelligence/ Machine Learning (AI/ML) Focused Open Topic

Scroll to Top