Integration of a fuzzy ART approach in a biologically inspired sensorimotor architecture

We present a biologically motivated computational model that is able to anticipate and evaluate multiple hypothetical sensorimotor sequences. Our model for anticipation based on cortical representations (MACOR) allows a completely parallel search at the neocortical level using assemblies of rate coded neurons for grouping, separation, and selection of sensorimotor sequences. For a vision-controlled local navigation of a mobile robot Khepera, we can demonstrate that our anticipative approach outperforms a reactive one. We also compare our explicitly planning approach with the implicitly planning Q-learning. Further, we show the advantages of ART as a realization of MACOR for real-world tasks.