Experiments with the subsumption architecture

Abstract

The Subsumption Architecture is a special case of behavior based control for robotics. Behavioral modules are added as "layers" with each layer performing a complete behavior. Higher level behaviors override lower level ones by taking control of their effectors or manipulating their internal states. The control layers are built up out offinite state machines connected by links that act essentially like wires. To test this architecture on a reasonably complex problem, a prototype airplane controller was developed. This controller jlies a simulated aircraft from take-off to landing and was run on a "C" based implementation of the subsumption architecture. Several lessons where learned from this effort. The subsumption architecture as currently defined is not sufiiently modular. A clean interface between different behaviors would be desirable. And finally, a more general relationship than strict hierarchy between high level and low level modules is required. None of these problems is insoluble within the behavior based approach but all must be solved i f realistic problems are to be dealt with. Some candidate solutions are given.

Integrated cognitive architectures: a survey

“…The subsumption architecture decomposes a problem in terms of the behaviours exhibited by the robots instead of the stages of information flowing within the controller as in a traditional AI design. Conversely, some researchers believed that traditional cognitive architectures would be useful for higher cognitive functions, while the subsumption architecture was only meant for reflexive tasks (Hartley and Pipitone 1991). Figure 6 depicts the subsumption architecture, comprising of a hierarchy of "simple" behaviour-based modules organized into layers (levels of competence).…”

“…All architectures discussed contain some form of learning, except the behaviour-based subsumption architecture, wherein all the behaviours produced in reaction to the environmental cues are pre-wired into the design of the architecture (Brooks 1999;Hartley and Pipitone 1991). Soar employs chunking as a general learning mechanism.…”

“…The objective of the game is to ensure a steady flow of goods from the producers to the consumers through the trading by the retailers and the truck agents played by the architecture (Butler et al 2001). Hartley and Pipitone (1991) have also attempted to use the subsumption architecture in airplane control.…”

Navigation technology of autonomous mobile robots in unknown environments

“…The deterministic methods have weaknesses in big search space and effectiveness. The navigation control in uncertainty environments has presented three architectures: the reactive architecture [1]- [3], the deliberative architecture [4] and the hybrid architecture [5]- [6]. The reactive architecture which commonly uses the potential field method has the trap problems of the V-shaped terrain and local minima [7].…”

Adaptive testing of controllers for autonomous vehicles

“…The autonomous controller, which is responsible for flying the air& and performing the landing on the carrier deck, was designed using a subsumption architecture approach [2]. The controller is composed of individual behaviors, operating at diikrent levels of abstraction, that communicate among themselves and together allow the aircraft to fly and to land.…”