Army AI Innovation Institute (A2I2)

Background
The U.S. Army Combat Capabilities Development Command - Army Research Laboratory (DEVCOM-ARL) is seeking innovative research and development solutions under the ARL Broad Agency Announcement (BAA) for Basic and Applied Scientific Research, specifically focusing on Artificial Intelligence (AI) and Machine Learning (ML).

The goal of this contract is to support the development of new technologies and translational research approaches in two primary areas: Command and Control (C2) and Large Pre-Trained Models (LPTM). This initiative aims to enhance decision-making capabilities in complex battlefield environments.

Work Details
The contract encompasses several core objectives across multiple thrust areas:
1. **Command and Control (C2)**:
- Generate synthetic data for ML models that accurately represent dynamic battlefield environments.
- Develop resilient perception algorithms to process multimodal data streams in real-time.
- Create small ML models for efficient AI capabilities in resource-constrained settings.
- Explore multimodal Gen AI models for enhanced decision support.
- Develop a distributed common operational picture integrating data from various sources.
- Compare current methods with LPTM-based methods to improve situational awareness.

2. **Battlespace Information and Knowledge Management**:
- Build a digital holistic representation of the battlespace, incorporating physical attributes and military doctrine.
- Create question-and-answer interfaces for efficient interaction with the battlespace representation.
- Develop techniques for insight extraction from complex datasets while mitigating risks associated with LPTMs.

3. **AI-Enabled Decision Dominance**:
- Develop decision aids for intelligence preparation of the battlefield.
- Create AI agents for COA generation that incorporate human intent and doctrinal rules.
- Conduct foundational research on LPTM reasoning capabilities for autonomous decision-making.

4. **Human-AI Integration**:
- Develop natural language interfaces for AI-aided decision making to facilitate communication between humans and AI systems.
- Create methodologies that incorporate human feedback into AI systems to improve performance and trust.

Period of Performance
The anticipated period of performance is 24 months, with the possibility of three additional option years (12 months each), depending on the proposals submitted.

Place of Performance
The geographic locations where the contract will be performed are not explicitly provided.