The Use of State in Intelligent Control

Much of the success of behavior based robotics can be attributed to the minimal representation approach proposed by Brooks in the mid-eightiesBrooks (1991). However, this has lead to a general “fear of state” which in turn leads to a neglect of the role of state in intelligent control. This paper discusses the need for propagating decision points in action selection in order to coordinate complex behavior. It also proposes a mechanism for doing this while maintaining reactive responsiveness to the real world. This mechanism involves the use of fixed action patterns and specialized behavior prioritization. The result is to focus control attention without prescribing a fixed course of action.