SBIR/STTR Research Award
Response to Deuce Coup: A Field tested multi-modal zero shot AI component decision engine
Award amount
$1,717,503Program and phase
SBIR · Phase II – Direct to Phase IIAward year
2024Award details
- Awarding agency
- Department of Defense
- Branch
- Defense Advanced Research Projects Agency
- Recipient company
- CHAOS INDUSTRIES, INC.
- Proposal award date
- 2023-12-18
- Contract end date
- 2025-04-17
- UEI
- P8SFGBHEHHF5
- Solicitation number
- 23.4
- Topic code
- HR0011SB20234-07
Research abstract
Chaos Industries offers a first of its kind mobile application, ōAnalyze & Design for Supply (ADFS)ö to help tackle the urgent global supply chain shortage issues affecting both military and commercial assetsŚtested against readily available commercial electronic componentsŚand to address the Deuce Coupe problem. ADFS not only identifies valuable and rare components to be reused in e-waste and junked systems, but assesses the lead time, availability risks, and cost of components of a system in real time and offers replacement suggestions. During the months leading up to the Deuce Coupe proposal, Chaos built a working mobile application meeting the Deuce Coupe Phase 1 feasibility criteria by leveraging ChaosÆ industry leading Design for X (DFX) AI platform, powered by our unified multi-modal, multi-task (M3T) model, an improvement on 4th gen generative transformer technology. The Chaos DFX team is led by four PhDs in edge AI, who recently sold their generative AI compiler product company and joined Chaos to help solve the DoD and DARPA mission; we are one of the few teams who have deployed generative AI products to the field in a mobile form factor. The DFX platform has been privately funded by millions in venture capital since 2022. The ADFS product we offer brings a rapid, dependable, and easily accessible solution to identify, evaluate, and salvage valuable mechanical and electronic components from discarded or damaged systems, thus enabling their reuse in operational systems. The ADFS system reports on exactly how rare, how long a lead time, and the projected cost of replacing these components would be. Our multi-modal generative AI model achieves 90% accuracy in these use cases.
Technology keywords
Not reported