Rodney A. Brooks introduced a new architecture for controlling mobile robots, emphasizing a layered control system that allows robots to operate at increasing levels of competence. The system uses asynchronous modules communicating over low-bandwidth channels, with each module being a simple computational machine. Higher-level layers can subsume lower-level roles, but lower levels continue functioning. This results in a robust and flexible control system. The system has been used to control a robot navigating unconstrained environments and is intended for office areas with mapping capabilities. The control system addresses multiple goals, sensors, robustness, and extensibility. It uses task-achieving behaviors as a primary decomposition, leading to a different architecture than traditional functional decomposition. The system is robust, allowing adaptation when sensors fail or environments change. It is extensible, with new layers added without altering existing ones. The system includes levels of competence, from avoiding contact with objects to planning and executing complex tasks. Layers of control correspond to these levels, with higher layers subsuming lower ones. Each layer is a finite state machine, communicating asynchronously. The system is designed to handle multiple sensors, with some data not needing central processing. It is robust, with lower layers continuing to function even when higher layers are active. The system has been implemented for levels 0 and 1, with level 2 under development. It has been tested on both simulated and physical robots, demonstrating effective navigation and obstacle avoidance. The system is extensible, with new processors added without significant changes to existing ones. The architecture is suitable for future expansion, with potential for custom chips and improved vision systems. The paper concludes that the layered control system offers a flexible and robust approach to mobile robot control, with implications for future robot design and implementation.Rodney A. Brooks introduced a new architecture for controlling mobile robots, emphasizing a layered control system that allows robots to operate at increasing levels of competence. The system uses asynchronous modules communicating over low-bandwidth channels, with each module being a simple computational machine. Higher-level layers can subsume lower-level roles, but lower levels continue functioning. This results in a robust and flexible control system. The system has been used to control a robot navigating unconstrained environments and is intended for office areas with mapping capabilities. The control system addresses multiple goals, sensors, robustness, and extensibility. It uses task-achieving behaviors as a primary decomposition, leading to a different architecture than traditional functional decomposition. The system is robust, allowing adaptation when sensors fail or environments change. It is extensible, with new layers added without altering existing ones. The system includes levels of competence, from avoiding contact with objects to planning and executing complex tasks. Layers of control correspond to these levels, with higher layers subsuming lower ones. Each layer is a finite state machine, communicating asynchronously. The system is designed to handle multiple sensors, with some data not needing central processing. It is robust, with lower layers continuing to function even when higher layers are active. The system has been implemented for levels 0 and 1, with level 2 under development. It has been tested on both simulated and physical robots, demonstrating effective navigation and obstacle avoidance. The system is extensible, with new processors added without significant changes to existing ones. The architecture is suitable for future expansion, with potential for custom chips and improved vision systems. The paper concludes that the layered control system offers a flexible and robust approach to mobile robot control, with implications for future robot design and implementation.