Charles River Analytics has been awarded by the Defense Advanced Research Projects Agency Strategic Technology Office a contract worth $9.9 million to develop automated strategies for planning complex operations.
The company said Monday it will work with Data Machines on Phase 2 of the project, which involves further development of the Meta-Reinforcement Learning System for Strategy and Tactics Assistance using Force-on-Force Forecasting, or MERLINS-STAFF. Phase 1 of the contract, worth $4 million, covered the development of a new neuro-symbolic artificial intelligence approach using strategic games to support the planning process.
What Is MERLINS-STAFF?
MERLINS-STAFF is a component of DARPA’s Strategic Chaos Engine for Planning, Tactics, Experimentation and Resilience, or SCEPTER, program. It aims to develop machine-generated strategies using neuro-symbolic AI to assist humans in planning operations and making complex decisions.
With the limitation of existing AI systems, DARPA aims to fully develop and utilize AI-enhanced planning through the SCEPTER program. The agency envisions a human-machine teaming system capable of adapting to different environments and developing statistically significant enhancements.
According to Michael Harradon, principal scientist at Charles River and principal investigator on MERLINS‑STAFF, real-time strategy computer games can be used to evaluate how new AI techniques perform in a controlled environment.
“In Phase 1, we achieved significant advancements, including playing longer games, reducing training time by 1000x, handling more complex entity controls and creating human-understandable plans,” said Harradon.