Smart and Autonomous Systems | 18 557

Opportunity Information: Apply for 18 557

The National Science Foundation (NSF) Smart and Autonomous Systems (SampAS) grant opportunity (Funding Opportunity Number 18-557) supports research aimed at building the next generation of Intelligent Physical Systems (IPS), meaning real-world machines and infrastructures that can sense, think, communicate, and act in physical environments. The central goal is robust, long-term autonomy: systems that can keep operating safely and effectively for extended periods with minimal or no human intervention, even when conditions are uncertain, changing, or completely unanticipated. The program is framed around the idea that today’s autonomous and semi-autonomous systems often work well only in narrowly defined settings, largely because they rely on pre-programmed behaviors or assumptions about what they will encounter. When they face something novel, many systems struggle to recognize they are outside their competence, let alone adjust their plans intelligently. SampAS is intended to push beyond that limitation so future IPS can understand their own boundaries, respond to surprises, and continue operating in a dependable way.

NSF highlights a wide range of example application domains to illustrate what counts as an IPS, including robotic platforms, self-driving vehicles, underwater exploration vehicles, and smart grids. What ties these examples together is the need to couple decision-making with physical action in complex environments, where sensor noise, incomplete information, shifting goals, and real-world constraints are unavoidable. In this context, SampAS emphasizes research that makes autonomy more resilient and self-sustaining, not merely better at executing a fixed set of tasks under ideal conditions. The program is therefore less about incremental performance improvements in a known benchmark environment and more about foundational capabilities that let systems remain reliable when the environment, the task, or the system’s own condition changes.

The opportunity is organized around four major research thrusts that describe the kinds of intelligence NSF wants IPS to develop: cognizant, taskable, adaptive, and ethical. Cognizant IPS are expected to have high-level self-awareness about what they can and cannot do. That includes being able to anticipate potential failures, detect when assumptions no longer hold, and re-plan accordingly rather than blindly continuing with an outdated strategy. Taskable IPS focus on communication and intent: they should be able to take high-level instructions that may be vague or incomplete and translate them into concrete action plans tailored to the specific context they are operating in. Adaptive IPS are expected to improve over time by learning from experience, whether that experience comes from the system’s own operation, from other systems, or from humans through instruction or observation. Ethical IPS incorporate societal and legal constraints into decision-making, meaning the system’s choices should account for rules, norms, and safety considerations rather than optimizing only for speed, efficiency, or task completion. Taken together, these four areas point to autonomy that is not only capable and flexible, but also accountable and aligned with real-world expectations for responsible behavior.

A key theme running through the solicitation is that achieving these capabilities will require knowledge-rich approaches. NSF explicitly calls out the need for a variety of representation and reasoning methods, such as semantic reasoning (working with structured meaning and relationships), probabilistic reasoning (handling uncertainty and noisy evidence), commonsense reasoning (making everyday inferences that are rarely spelled out), and meta-reasoning (reasoning about the system’s own reasoning processes, confidence, and limits). In practical terms, this signals interest in systems that can combine learning with explicit reasoning, maintain internal models of competence and risk, and justify or revise decisions as circumstances evolve. The expectation is not simply that a system can learn patterns, but that it can use structured knowledge to decide when to ask for help, when to change tactics, and how to stay within constraints.

From an administrative standpoint, this is a discretionary NSF research grant in the science and technology / research and development category (CFDA 47.070). The listed award ceiling is $1,000,000, with an expected 25 awards under this opportunity. The opportunity was created on May 2, 2018, with an original closing date of July 31, 2018. Eligibility is listed as “Others,” with details referenced in an additional eligibility section in the full solicitation. Overall, the grant is designed to fund research that advances core methods and system-level designs for autonomous physical systems that can operate longer, handle novelty better, accept more natural and flexible tasking, learn and adapt over time, and make decisions that reflect ethical, legal, and societal constraints.

• The National Science Foundation in the science and technology and other research and development sector is offering a public funding opportunity titled "Smart and Autonomous Systems" and is now available to receive applicants.
• Interested and eligible applicants and submit their applications by referencing the CFDA number(s): 47.070.
• This funding opportunity was created on May 02, 2018.
• Applicants must submit their applications by Jul 31, 2018. (Agency may still review applications by suitable applicants for the remaining/unused allocated funding in 2026.)
• Each selected applicant is eligible to receive up to $1,000,000.00 in funding.
• The number of recipients for this funding is limited to 25 candidate(s).
• Eligible applicants include: Others (see text field entitled Additional Information on Eligibility for clarification).

Apply for 18 557

[Watch] Creating a grant proposal using the step-by-step wizard inside the applicant portal: