自主发育智能机器人体系结构研究

RESEARCH ON THE ARCHITECTURE OF AUTONOMOUS DEVELOPMENTAL ROBOT

传统的机器人系统范式分类已经无法将新出现的方法和理论纳入其中。为此,首先从认知的角度重新对机器人的范式进行分类。新的范式分类涵盖了传统的系统范式,明确了自主发育在机器人系统范式中的地位。在此基础上,提出了自主发育智能机器人体系结构。该结构只需利用基本的感知能力和行动能力,分别利用感知发育模块、认知发育模块和行为发育模块实现自主感知分类、时空经验知识以及反应式行为的逐层发育。各发育模块之间互相依赖并可以同时学习,具有实时的自主发育能力。

Traditional robot classification paradigm can no longer cover new emerging methods and theories. For this reason, the paper firstly reclassifies the paradigm of robot architecture from the cognitive point of view. New paradigm classification not only covers the traditional paradigm, but also specifies the importance of autonomous development in the paradigm of robot architecture. On this basis, autonomous developmental robot architecture is proposed. The architecture only needs such fundamental capabilities as perception and action in order to achieve the hierarchical development of autonomous perception classification, spatio-temporal experience and reactive behavior with development modules for perception, cognitive and behavior separately. Development modules are interdependent and can learn synchronously so as to possess the capability of real-time autonomous development.