Attention in Autonomous Robotic Visual Search

One important element of autonomy for mobile robots, both terrestrial and planetary, is the ability to search for and find targets of interest whether they be important for navigation, for science missions, for localization or for manipulation. In this paper, we introduce a novel method of autonomous visual search that exploits the use of attention in the form of a saliency map that is used to enhance the probability distribution of which areas to look next, increasing the utility of spatial volumes, where objects consistent with the target’s visual saliency are observed. Experimental results on a practical mobile robot are presented, which show that our proposed model improves the process of visual search in terms of reducing the time and number of actions to be performed to complete the process.