Develop methods for the validation and deployment of AI-powered robots in manufacturing applications
Active Cooperative Agreement Awarded Last action
- Award ID
- 70NANB24H314
- Recipient
-
Carnegie-Mellon University UEI U3NKNFLNQ613
- Awarding agency
-
National Institute of Standards and Technology
Department of Commerce
- Obligated
- $198,000
Award facts
- Assistance listing
- 11.609 Measurement and Engineering Research and Standards
- Completion date 69% complete
- Originally
- Place of performance
- Pittsburgh, PA 15213, USA
Award details come from USAspending.gov.
Description
- Purpose
- The purpose of this grant is to develop methods for the validation and deployment of AI-powered robots in manufacturing applications. The research will enable the benchmarks and standardized evaluation metrics for robot AI models to reduce the adoption risk in real applications.
- Activities
- The proposal will perform the following three activities:1) robotic manufacturing testbeds via generative simulation ? They will develop an AI robotic information system that involves creating a generative simulation framework that enables manufacturers the ability to automatically generate robotic manufacturing test-beds with thousands of environmental variations. They will use large language models (llms) to parse language descriptions of the environment to extract the desired objects, spatial layouts, and tasks goals. They will also use the vision language model (vlms) to extract information from images or videos, or a combination of both. Object assets required for the simulation can either be generated using text-to-3D object generative models or retrieved from existing object datasets.2) automatic generation of environmental and task variations for learned robot model validation ? They will explore methods to generate realistic and plausible environment variations within manufacturing environments using generative simulation. These variations may include changes in object size, shape, visual texture, and physical properties, as well as the initial configuration of robots and objects (e.g., positions and joint angle configurations), task objectives (e.g., target placement locations of objects), and environmental factors such as lighting conditions. They will prompt an llms to suggest these variations, ensuring they are realistic and contextually appropriate. All of the generated variations can be used to validate the robustness of existing robot learning models, or to train new models to be robust to such variations. 3) automatic analysis of results and patterns - they will develop a method to automatically analyze the performance of robot AI models in the generated variations of the manufacturing environments. They will develop automatic techniques to identify the failures in learned robotic manufacturing control models, cluster the failure cases, and explain the possible reasons for the failures, which can help identify common regions where a learned robot model fails to generalize to realistic variations in a manufacturing setup and provide guidance on prioritizing areas for model improvement.
- Expected outcomes
- The deployment of the proposed methodologies will address the key challenge of the validation and deployment of AI-powered robots in manufacturing applications. The success of the project will enable the development of benchmarks and standardized evaluation metrics for robot AI models in industry. IT will greatly benefit the industry by increasing reliability to introduce AI-driven automation into manufacturing facilities at a greater scale, leading to increased efficiency, productivity, and innovation in the industry.
- Intended beneficiaries
- Robot end users, robot vendors, robot systems integrators, manufacturers, U.S. Industrysubrecipient activities: the recipient does not intend to subaward funds.
Transactions
| Action | Obligated | Date |
|---|---|---|
| Amendment | $0 | |
| Original award | $198,000 |
Quick answers
Who received award 70NANB24H314?
70NANB24H314 was awarded to Carnegie-Mellon University by National Institute of Standards and Technology (Department of Commerce) on October 1, 2024.
How much is award 70NANB24H314 worth?
$198,000 has been obligated on 70NANB24H314.
On GovTribe
See this award's full record
- Every transaction and subaward
- The recipient's other awards and teaming partners
- Similar grant awards
- Ask GovTribe AI about this award