Richard Fikes and Nils Nilsson Introduce STRIPS

If you started from first principles, could you solve the world with only logic? Yeah probably not.

Remember how before we said (the split after Perceptrons) that this period in AI history was a boom time for symbolic, rules-based AI (and a subsequent winter for cybernetic, learning, connectionist AI)? Well, here is another example of "experts" trying to get expert systems to work: STRIPS.

Richard Fikes and Nils Nilsson of SRI presented STRIPS, the Stanford Research Institute Problem Solver, at the International Joint Conference on Artificial Intelligence in London in September 1971.

Built for SRI’s robot research, it described the world in first-order logic and each action as an operator with a precondition, a delete list of facts that may stop being true and an add list of facts that become true. It then searched for a sequence of operators reaching the goal. Asked to gather three boxes in one place, it returned four steps: go, push, go, push.

Unlike others from this period, STRIPS had lasting power. The Planning Domain Definition Language, a community standard since 1998, has STRIPS at its core.