This paper proposes an algorithm to estimate appropriate or novel content for groups of people who know each other such as friends, couples, and families. To achieve high recommendation accuracy, we focus on "Groupality", the entity or entities that characterize groups such as the tendency of content selection and the relationships among group members. Our algorithm calculates recommendation scores using a feature space that consists of the behavioral tendency of a group and the power balance among group members based on individual preference and the behavioral history of group. After gathering the behavioral history of subject groups when watching TV, we verify that our proposed algorithm can recommend appropriate content, and find novel content. Evaluations show that our proposal achieves higher performance than existing methods.

A multiple-sensor-enhanced robot collaboration system has been developed for manufacturing applications, The system consists of basic robot modules, such as for locating and trading. These modules have laser range finders and perform hybrid manufacturing tasks by working together as an advanced robotic cell. A hybrid task with uncertainties can be described by the task velocity and task error derived from the sensory and motion information on the robot modules. The modules send their bask error and waiting time to each other via a collaboration network. They also modify their given motion sets in real time based on fuzzy rules in order to improve the accuracy and efficiency of the system. An experimental study of multi-station welding showed that this collaboration results in a satisfactory trade-off between accuracy and efficiency.

We have developed a highly reliable method of detecting contact-state transitions in real time. In the detection algorithm, feature variables, which fluctuate abruptly when a contact state transition occurs, are selected. The differential of the feature trajectory is monitored to distinguish the transition region from the stable region. To confirm that a state transition has actually occurred in the region, the duration of the region is measured. With this method, a contact-state transition can be detected with a 40-msec delay. We also have developed an experimental manipulation system, in which motion execution management and motion monitoring and execution modules operate independently. Using transputers and a six-degree-of-freedom direct-drive manipulator, a square-peg insertion task is carried out to evaluate the effectiveness of the detection method in an actual environment. The experimental results show that contact-state transitions can be accurately detected and that motion commands can be quickly switched by using the detected transition information.

New methods for teaching and controlling contour-tracking tasks based on contact sensing are developed to improve robotic machining. Accurate contact point detection is introduced to estimating a work contour path with high reliability. These teaching methods reduce the effort for task teaching and improve adaptability to handle changes in task specifications without re-teaching. Real-time contour tracking control using this contact sensing to suppress work uncertainties in process is presented. Experimental results for practical grinding tasks using a 6 DOF manipulator validate our methods.>