The Community for Technology Leaders
Green Image
We introduce mechanisms for attention control and pattern categorization as the basis for cognition in a humanoid robot. We hope to identify principles of human-cognitive organization that might depend in part on the opportunities provided in the human morphology-hence, our interest in humanoid form factors. Humanoid robots promise to lead us toward more effective and informative interactions between man and robot devices. The proposed mechanisms support complex, cooperative, active, and multimodal sensory systems. We describe the hardware and scientific perspective underlying the UMass humanoid torso-affectionately named "Magilla" by virtue of its gorilla-like appearance. We motivate a framework for cognitive integration of multiple sensory and motor modalities and provide some preliminary results of the framework in the attentional control of Magilla's articulated head. As a practical result, the robot is able to monitor an extended region of its environment. This involves constructing salience maps for controlling attentional shifts, an efficient feature extraction architecture, and pattern categorization. The system constructs attentional maps incrementally and maintains consistency in the maps given current perceptions of the environment. We present demonstrations on a variety of monitoring tasks.
attention control, multimodal stereognosis, pattern categorization, robot cognition
Andrew H. Fagg, Luiz-Marcos Garcia, Roderic A. Grupen, David S. Wheeler, Antonio A.F. Oliveira, "Tracing Patterns and Attention: Humanoid Robot Cognition", IEEE Intelligent Systems, vol. 15, no. , pp. 70-77, July/August 2000, doi:10.1109/5254.867915
89 ms
(Ver 3.3 (11022016))