Learning Introspective Control (LINC)

The Defense Advanced Research Projects Agency seeks proposals to develop machine learning technologies enabling adaptive control of physical systems through its Learning Introspective Control program. Proposals are solicited for Technical Area 1 to use onboard sensors and actuators to reconstitute control laws in real-time responding to unforeseen events while maintaining stability and operator guidance. Proposals are due by November 24, 2021 and must address technical approach, costs, transition plans, and team expertise across three program phases over four years involving exercises and end-of-phase testing using government equipment.

There is no set-aside designation. The solicitation involves research in areas such as adaptive control, generative adversarial networks, autonomous systems, and operator cueing. Multiple awards are anticipated totaling up to $24 million for Phase 0 through Phase 2 focusing on control of modified military ground robots and vehicles. Proposals must follow formatting requirements and will be evaluated based on technical merit, transition potential, and cost considerations.