Control of Mobile Robots Using the Soar Cognitive Architecture

Abstract

This paper describes the development of a system that uses computational psychology (the Soar cognitive architecture) for the control of unmanned vehicles. A multithreaded software system written using Java and integrated with the Soar cognitive architecture has been implemented on two types of mobile robots. Soar can be used as a general purpose robotic intelligence system and can handle a wide variety of sensor inputs and motor-control outputs. The use of existing computational psychology methods (such as Soar) may be a more efficient approach to developing robotic software, rather than developing new software for every new unmanned vehicle application. Results from the application of this software system (named the cognitive robotic system, or CRS) to a practical unmanned ground vehicle mission, navigating to a target GPS location while avoiding obstacles, are described. The CRS has been designed so that its capabilities can be expanded in the future through the inclusion of additional sensors and motors, additional software systems, and more sophisticated Soar agents.

Deep Blue Cannot Play Checkers: The Need for Generalized Intelligence for Mobile Robots

“…The sensor inputs and motor control outputs are controlled in Java while the symbolic data is stored in Soar. CRS has been used on both wheeled and legged robots [4]. It has also been used with several types of sensor input systems (vision, sonar, GPS, compass, touch, etc.).…”

“…Examples of these are Soar [14], ACT-R [15], and EPIC [16]. Some of these have been implemented on mobile robots [4,17].…”

“…Current mobile robots have very little intelligence or autonomy. Intelligence would allow the robots to perceive their environment, plan their activities, solve problems, and learn from experience [4]. Autonomy would allow them to survive in a real-world environment without external control.…”

The Requirements and Possibilities of Creating Conscious Systems

“…This system uses a cognitive architecture (based on ACT/R) for decision making but allows for sub-symbolic processing, such as neural networks, for processing sensor data. We have also used the Soar cognitive architecture coupled to sensor inputs and motor outputs [50], as shown in Figure 6. These hybrid approaches will be needed in the near term to build intelligent systems.…”

“…Even cognitive architectures (e.g. [46], Soar [47,48], SS-RICS [48], and ACT/R [49]), which have been implemented on mobile robots ( [50], [48]), are not even close to human intelligence and power and have only rudimentary learning ability. The IEEE Computational Intelligence Society is more focused on sub-symbolic processes such as neural networks, genetic algorithms, and fuzzy logic, which may lead to large-scale intelligent systems.…”