Automated Discovery for Design and Control of Turbulent Systems (AutoDIDACTS)

The Automated Discovery for Design and Control of Turbulent Systems (AutoDIDACTS) program is a federal funding opportunity issued by the Defense Advanced Research Projects Agency (DARPA), specifically through its Defense Sciences Office. DARPA is a research and development agency within the U.S. Department of Defense that focuses on advancing breakthrough technologies to support national security. This Broad Agency Announcement outlines a research initiative aimed at transforming aeronautical design and turbulence control through data-driven discovery of generalizable principles and surrogate models. The purpose of the AutoDIDACTS program is to enable the development of data-informed methodologies that can generate predictive, generalizable macroscale design principles for turbulent systems. The program seeks to overcome longstanding limitations in computational and experimental aerodynamics by combining high-quality experimental data with advanced machine learning and modeling approaches. Rather than relying on computationally intensive simulations or narrow empirical models, the program emphasizes the automated discovery of functional relationships that can guide design exploration and optimization across a wide range of flow regimes. Funding under this program supports research efforts that integrate three key capabilities: development of compact and tunable turbulence simulators, acquisition of high-resolution experimental data, and application of AI-based techniques to derive predictive models. The scope includes both the creation of new experimental platforms and the development of algorithms capable of extracting meaningful design principles. The program explicitly excludes incremental improvements to existing technologies and instead prioritizes innovative approaches that could lead to transformative advances. The program is structured as a 48-month effort divided into two phases. Phase I, which is the focus of this solicitation, spans 24 months and involves developing experimental data systems and initial models. Phase II, also 24 months, will involve demonstration of solutions to challenge problems defined in a future solicitation. During Phase I, performers are expected to demonstrate model accuracy exceeding 95 percent and generalizability across more than 100 times the parameter space. Regular program reviews and an intermediate demonstration at the end of the first year are required to assess progress. Eligibility for this opportunity is broad, allowing participation from U.S. and non-U.S. organizations, including academic institutions, private companies, nonprofits, and government-affiliated research entities, provided they comply with applicable regulations. Multiple award instruments may be used, including procurement contracts, cooperative agreements, and other transaction agreements. While cost sharing is not universally required, it may apply depending on the award type. Proposals must include detailed technical plans, risk mitigation strategies, and a clear path toward meeting program objectives. The application process includes an optional abstract submission followed by a full proposal. Proposals must be submitted through designated federal systems depending on the award type, such as the Broad Agency Announcement Tool or Grants.gov. Required components include technical and management volumes, cost proposals, and supporting documentation such as biosketches and current support disclosures. Evaluation criteria emphasize scientific merit, relevance to DARPA’s mission, and cost realism. Key deadlines include an initial announcement on April 1, 2026, an abstract due date of April 15, 2026, and a full proposal deadline of May 28, 2026, at 4:00 PM Eastern Time. Questions were accepted until May 15, 2026. Awards are anticipated to begin around September 2026. Program contacts include a technical program manager and a central program email address for inquiries.

Focus on highly innovative approaches rather than incremental improvements; clearly demonstrate generalizability and model accuracy; include strong risk mitigation strategies