CAJ | 학술논문

针对目前智能移动机器人在未知环境中学习遇到的如学习主动性、实时性差，无法在线积累学习的知识和经验等问题，受心理学中内部动机的启发，提出一种内部动机驱动的移动机器人未知环境在线自主学习方法，在一定程度上弥补目前该领域存在的一些问题。该方法通过在移动机器人Q学习的框架下，将奖励机制用基于心理学启发的内部动机取代，提高其对于未知环境的学习主动性，同时，采用增量自组织神经网络代替经典Q学习中的查找表，实现输入输出空间的映射，使得机器人能够在线增量地学习未知环境。实验结果表明，通过内部动机驱动的方法，移动机器人对于未知环境的学习主动性得到了提高，智能程度有了明显改进。
침대목전지능이동궤기인재미지배경중학습우도적여학습주동성、실시성차，무법재선적루학습적지식화경험등문제，수심이학중내부동궤적계발，제출일충내부동궤구동적이동궤기인미지배경재선자주학습방법，재일정정도상미보목전해영역존재적일사문제。해방법통과재이동궤기인Q학습적광가하，장장려궤제용기우심이학계발적내부동궤취대，제고기대우미지배경적학습주동성，동시，채용증량자조직신경망락대체경전Q학습중적사조표，실현수입수출공간적영사，사득궤기인능구재선증량지학습미지배경。실험결과표명，통과내부동궤구동적방법，이동궤기인대우미지배경적학습주동성득도료제고，지능정도유료명현개진。
In order to solve the problems in intelligent mobile robot learning in unknown environment such as low activity and real-time performance, unable to accumulate learning experience. Inspired by internal motivation in psychology, a method called internal motivation driven mobile robot unknown environment learning algorithm is proposed to make up the problems above to some extent. The proposed method architecture is Q-learning, and the reward is modified by the internal motivation inspired by psychology, to raise the performance of learning activity. The look-up table in Q-leaning is also replaced by incremental self-organized network, to realize the map from input to output. So the robot is able to incre-mentally learning unknown environment in a online manner. The experiment result shows that by the internal motivation driven method, the learning activity for learning the unknown environment is raised, the intelligence of the robot is also improved.