Smart Stereovision Based Gaze Control for Navigation in Low-Feature Unknown Indoor Environments

State-of-the-art mobile robot navigation systems usually consider Gaze control and locomotion planning to be independent modules. We present an integrated gaze control strategy that maximizes the collection of relevant 3D measurements of environment. The presented work results in a more reliable environment representation compared to state-of-the-art approaches for unknown, dynamic indoor environment. In our particular approach, Grid based mapping is employed for path planning purposes. Such an approach demands a sufficient amount of 3D information from stereovision sensors for effective navigation, especially in environments with relatively less number of features such as plain walls and floors. We conduct several experiments in such environments and compare our results with contemporary stereovision based mapping and path planning approaches. We present results of multiple experiments, to show the effectiveness of our system to be able to navigate in low-feature environments using solely, stereovision based gaze control strategy for path planning.