论布鲁克斯的无表征智能理论

On Rodney Brooks' Theory of Intelligence without Representation

无表征智能是美国著名机器人制造专家罗德尼.布鲁克斯(Rodney Brooks)提出的一种智能理论。布鲁克斯认为,传统人工智能在建造机器人时,采取的是"感觉-建模-计划-行动"框架,这种框架在表征问题上遇到了巨大的困难。实际上,机器人的行动方式完全不必如此复杂,只需要两个步骤就可以:即感觉,然后行动。我们在建造机器人时,完全可以放弃表征范式所要求的那些复杂的计划、映射和认知等表征内容,而采取基于行为的包容结构。在包容理论指导下,布鲁克斯制造了"艾伦"、"赫伯特"、"成吉斯"、"考格"等机器人。布鲁克斯的理论取得了一定的成功,但也遇到一些新的困难。

Intelligence without representation is a theory of intelligence proposed by American robot maker Rodney Brooks. Brooks holds makers of traditional artificial intelligence use scheme of sense-model-plan-actionas their theoretical base. But this scheme has enormous difficulties because of representation. Brooks believes that the mode of robot action is not necessarily too complex. It＇s only needs two steps for its action： sense and act. We can completely abandon those complex contents of planning, mapping and cognizing which are demanded by representational paradigm, and take behavior-based subsumption architecture as a new robot building model. Directing by this theory, Brooks built robots such as Alan, Herbert, Genghis, and Cog, etc.. Brooks gains many successes, but he also meets some difficulties.