Unsupervised learning to recognize environments from behavior sequences in a mobile robot

摘要: We describe the development of a mobile robot which does unsupervised learning for recognizing environments from behavior sequences. Most studies on an environment have tried to build precise geometric maps with high sensitive and global sensors. However such information may not be obtained in real environments. Furthermore unsupervised-learning is necessary recognition unknown without help teacher. Thus we attempt recognize low sensitivity local The behavior-based wall-following enclosures. Then sequences behaviors executed each enclosure are transformed into input vectors self-organizing network. Learning teacher done, becomes able identify Moreover developed method independent start point using partial sequence. fully implemented system robot, made experiments evaluating ability. As result, found out that was done well our adaptive noisy