CPS: Creative Problem Solver

Background:

In the military, autonomous systems are increasingly available, but have brittle intelligence and narrow applications, with no ability to operate in ways not originally “engineered in.” CPS will enable broader uses and more robust operations, with autonomous systems operating with imprecise models, fighting through failure, and intelligently reconfiguring themselves and their assets to pursue mission goals.

Commercial applications will create autonomous systems capable of finding novel solutions for remote or highly uncertain applications. Mixed-initiative versions could also automate portions of early-phase system design, where qualitative modeling guides low-cost early decision-making.

SIFT Approach:

CPS solves problems using analogical reasoning, configuration planning over qualitative physics, hybrid systems planning, and active experimentation.

Benefits:

  • More robust (less brittle) in novel situations than existing approaches: CPS will succeed in off-nominal scenarios by creating new solution methods.
  • Avoids over-precise models and resulting brittleness of existing autonomous planning solutions.