Grant Details

Grants Office Grantwriting service fee is currently unavailable for this grant
Get more information on grantwriting

Authority

Summary

History of Funding

A map of recent awards that have been made through this project can be seen at https://www.nsf.gov/awards/award_visualization.jsp?org=NSF&pims_id=505325&ProgEleCode=039Y&from=fund.

Additional Information

The following are types of research that the S&AS program anticipates supporting. Proposals need not address all of these example themes, but should be of sufficient scope that they make meaningful progress towards the objectives of the program:

• Cognizant: IPS should exhibit high-level awareness beyond primitive actions, in support of persistent and long-term autonomy. IPS should be aware of their own capabilities, understand their limitations, anticipate possible failures, undertake contingency planning, and recognize when they are operating incorrectly and adapt accordingly. This high-level awareness includes a variety of IPS capabilities, such as: (a) being capable of operating autonomously over extended periods of time, with minimal or no human supervision; and (b) acting in a manner that is understandable by their human collaborators, and being capable of explaining themselves and asking for help, when needed.
• Taskable: IPS should have the ability to accept high-level, possibly vague, instructions from humans and be capable of translating such goal-oriented directives into a suitable plan for sensing, reasoning, communicating, and acting through the underlying system architecture. Instruction modes for the IPS may include natural language, gestures, and end-user programming techniques, such as sketches and multi-modal dialog systems. IPS should be able to plan and execute a wide range of diverse tasks, in a wide variety of applications. IPS may take the initiative to achieve necessary or important tasks, should be interruptible, and should be capable of performing multiple tasks concurrently.
• Adaptable: IPS should be capable of optimizing, reconfiguring, and repairing autonomously. They should learn to improve their behaviors over time, including acquiring, modifying, and transforming their skills by augmenting their knowledge on how to perform tasks. They should also autonomously improve the models they use to perceive, plan, and act. These improvements can be the result of past experiences of the IPS when interacting with the environment; observations of how other entities, such as other IPS or humans, interact with the environment; or instructions received by the IPS on how to perform tasks. IPS should learn to perform new types of tasks with little or no human intervention needed to modify their designs or configurations. As a result, IPS are expected to continually adjust their behaviors and be capable of adapting to the particular contexts in which they are operating, even when specifics of those contexts were not initially modeled.
• Ethical: IPS should adhere to an ethical system of societal and legal rules. IPS are expected to be capable of ethical reasoning, such as incorporating societal values into their reasoning. When reasoning about how their actions might affect the environment, IPS should consider not only how to achieve their tasks but also the consequences of their actions, from an ethical perspective. IPS should recognize moral imperatives, and avoid commands and actions that violate those dictates. Ethical IPS should recognize when multiple moral imperatives are in conflict and correctly apply ethical decision making.
• Knowledge-Rich: IPS should be able to represent and reason with knowledge at multiple levels of abstraction. IPS should be capable of quantitative and qualitative reasoning, leading to facility with high-level semantic concepts. IPS may reason using multiple models, such as symbolic, ontological, probabilistic, mixed, and commonsense reasoning models. IPS should be capable of meta-reasoning, introspectively reasoning about their own reasoning capabilities and limitations.

Contacts