Subsumption architecture

Subsumption architecture is a robotics and artificial intelligence (AI) control architecture that was introduced by Rodney Brooks in the mid-1980s. It is a method for designing complex systems by breaking them down into simpler, behavior-based modules that operate independently and in parallel.

Overview

Subsumption architecture is based on the idea that intelligent behavior can emerge from the interaction of simple behaviors. Each behavior module is designed to achieve a specific goal and operates independently of other modules. These modules are organized in layers, with higher layers able to subsume or inhibit the actions of lower layers. This hierarchical structure allows for the creation of complex behaviors from simple building blocks.

Key Concepts

• Behavior-based systems: In subsumption architecture, the system is divided into a set of behaviors, each responsible for a specific task. These behaviors are typically simple and reactive, responding directly to sensory inputs.
• Layered control: The architecture is organized into layers, with each layer representing a different level of abstraction. Lower layers handle basic tasks, while higher layers manage more complex behaviors. Higher layers can override or inhibit the actions of lower layers.
• Parallel processing: All behavior modules operate in parallel, allowing the system to respond quickly to changes in the environment.
• Emergent behavior: Complex behaviors emerge from the interaction of simpler behaviors, rather than being explicitly programmed.

Applications

Subsumption architecture has been used in a variety of robotics applications, including mobile robots, autonomous vehicles, and robotic vacuum cleaners. It is particularly well-suited for tasks that require real-time responses to dynamic environments.

Advantages and Disadvantages

Advantages

• Simplicity: The modular design makes it easier to develop and debug individual behaviors.
• Robustness: The system can continue to function even if some modules fail, as other modules can compensate.
• Scalability: New behaviors can be added without significantly altering the existing system.

Disadvantages

• Limited planning: Subsumption architecture is primarily reactive and does not inherently support long-term planning or complex decision-making.
• Coordination challenges: Ensuring that multiple behaviors work together harmoniously can be difficult.

See Also

• Behavior-based robotics
• Reactive planning
• Rodney Brooks
• Artificial intelligence
• Mobile robot