- New project funded by the Defense Advanced Research Projects Agency (DARPA) will show how disparate AI agents of machines can intuitively work together to achieve a common mission objective
- Project will leverage GE Research’s Sherlock System, a unique dynamic sensing and controller platform developed with DARPA to experiment with a network of independent agents in various evolving scenarios
- A key goal is to mimic a human’s intuition and innate ability to work collaboratively with limited verbal and non-verbal cues from others
NISKAYUNA, NY – September 17, 2020 – Could independent industrial machines or systems intuitively figure out how to work together and achieve a common mission or goal? That’s what a team of scientists from GE Research and the University of California, Los Angeles (UCLA), will aim to demonstrate as part of a project supported through DARPA’s Context Reason for Autonomous Teaming (CREATE) program.
“Much like a team of basketball players work together on the court to score a basket, we’re aiming to create AI agents that give industrial machines and systems that same intuitive and innate human ability to react to each other during emergent, or evolving situations to achieve shared goals,” said Peter Tu, Chief Scientist for Artificial Intelligence at GE Research. “This will involve a whole new level of AI capabilities, in which AI agents can interpret, react or respond on their own without being prompted by direct communications or commands.”
Tu explained that performances in team sports illustrate very well the ability of human beings to adapt and work well with others to achieve victory. And the level of cooperation, he says, can vary depending on how long or often a group of people has played together.
Tu said, “It’s amazing to see how well one player can anticipate another’s movement or action when making or receiving a pass. But what’s even more fascinating is how two or more people can read or sense each other’s actions even when they only have played together for a short time. It’s the essence of this natural reaction to emergent behavior that we are looking to capture with the AI agents we’re developing for machines.”
Caption: Through DARPA's CREATE program, GE researchers will explore novel methods for how teams of autonomous AI agents can learn to cooperate in the absence of centralized command and with minimal communication. From an industrial perspective, this could benefit assets such as wind turbines that must operate in a cooperative fashion to achieve the best possible performance and efficiency.
GE and UCLA will conduct in-depth experiments on GE Research’s Sherlock System, a dynamic sensing platform built with DARPA that that is enabled by AI and computer vision. This platform can see and interpret various visual cues and behavioral expressions from a group of individuals to assess the overall mood of a crowd. Through the CREATE project, the research team hopes to expand these capabilities to show how heterogeneous groups of individuals can not only see and interpret visual cues and expressions but also act on them to achieve a commonly identified goal.
DARPA initiated the CREATE Program to accelerate new developments that enable intuitive machine-to-machine capabilities that are contextually aware and have the ability to learn and adapt to unexpected missions and achieve operational success. The key is machines being able to learn, adapt and act without any centralized direction or control.
About GE Research
GE Research is GE’s innovation powerhouse where research meets reality. We are a world-class team of scientific, engineering and marketing minds working at the intersection of physics and markets, physical and digital technologies, and across a broad set of industries to deliver world-changing innovations and capabilities for our customers. To learn more, visit our website at https://www.ge.com/research/.