Ryo Yoshino's research while affiliated with Ritsumeikan University and other places

In this paper, we propose an active perception method for recognizing object categories based on the multimodal hierarchical Dirichlet process (MHDP). The MHDP enables a robot to form object categories using multimodal information, e.g., visual, auditory, and haptic information, which can be observed by performing actions on an object. However, performing many actions on a target object requires a long time. In a real-time scenario, i.e., when the time is limited, the robot has to determine the set of actions that is most effective for recognizing a target object. We propose an active perception for MHDP method that uses the information gain (IG) maximization criterion and lazy greedy algorithm. We show that the IG maximization criterion is optimal in the sense that the criterion is equivalent to a minimization of the expected Kullback–Leibler divergence between a final recognition state and the recognition state after the next set of actions. However, a straightforward calculation of IG is practically impossible. Therefore, we derive a Monte Carlo approximation method for IG by making use of a property of the MHDP. We also show that the IG has submodular and non-decreasing properties as a set function because of the structure of the graphical model of the MHDP. Therefore, the IG maximization problem is reduced to a submodular maximization problem. This means that greedy and lazy greedy algorithms are effective and have a theoretical justification for their performance. We conducted an experiment using an upper-torso humanoid robot and a second one using synthetic data. The experimental results show that the method enables the robot to select a set of actions that allow it to recognize target objects quickly and accurately. The numerical experiment using the synthetic data shows that the proposed method can work appropriately even when the number of actions is large and a set of target objects involves objects categorized into multiple classes. The results support our theoretical outcomes.

In this paper, we propose an optimal active perception method for recognizing multimodal object categories. Multimodal categorization methods enable a robot to form several multimodal categories through interaction with daily objects autonomously. In most settings, the robot has to obtain all of the modality information when it attempts to recognize a new target object. However, even though a robot obtains visual information at a distance, it cannot obtain haptic and auditory information without taking action on the object. The robot has to determine its next action to obtain information about the object to recognize it. We propose an action selection method for multimodal object category recognition on the basis of the multimodal hierarchical Dirichlet process (MHDP) and information gain criterion. We also prove its optimality from the viewpoint of the Kullback--Leibler divergence between a final recognition state and a current recognition state. In addition, we show that the information gain has submodularity owing to the graphical model of the MHDP. On the basis of the submodular property of the information gain criterion, we propose sequential action selection methods, a greedy algorithm, and a lazy greedy algorithm. We conduct an experiment using an upper-torso humanoid robot and show that the method enables the robot to select actions actively and recognize target objects efficiently.