Visual novelty based internally motivated Q-learning for mobile robot scene learning and recognition

In the intelligent robot research field, the traditional machine learning paradigm is commonly used, which causes problems of low learning initiative, lack of adaptability with uncertainty and bad expansibility of knowledge and ability. According to the new research direction called cognitive development learning, an incremental and autonomous visual learning algorithm based on internally motivated Q-learning is proposed for mobile robot scene learning and recognition. The visual novelty is calculated by online PCA and is considered as the internal motivation for the Q-learning robot. The active learning and accumulation of knowledge is implemented in the form of updating PCA subspace, which is guided by internally motivated Q-learning. By equipped with the proposed algorithm, robot makes next learning decision by judging the novelty between learned knowledge and what is seen now. Experimental results show that the algorithm has the ability of autonomous exploring and learning, actively guiding robot to learning new knowledge, acquiring knowledge and developing intelligence in a online and incremental manner.