Mihoko Otake

Human body contour data based activity recognition

This research work is aimed to develop autonomous bio-monitoring mobile robots, which are capable of tracking and measuring patients motions, recognizing the patients behavior based on observation data, and providing calling for medical personnel in emergency situations in home environment. The robots to be developed will bring about cost-effective, safe and easier at-home rehabilitation to most motor-function impaired patients (MIPs). In our previous research, a full framework was established towards this research goal. In this research, we aimed at improving the human activity recognition by using contour data of the tracked human subject extracted from the depth images as the signal source, instead of the lower limb joint angle data used in the previous research, which are more likely to be affected by the motion of the robot and human subjects. Several geometric parameters, such as, the ratio of height to weight of the tracked human subject, and distance (pixels) between centroid points of upper and lower parts of human body, were calculated from the contour data, and used as the features for the activity recognition. A Hidden Markov Model (HMM) is employed to classify different human activities from the features. Experimental results showed that the human activity recognition could be achieved with a high correct rate.

A system that assists group conversation of older adults by evaluating speech duration and facial expression of each participant during conversation

In super aged society, system that assists social activities of older adults is needed for cognitive enhancement. Group conversation is one of the social activities. One of the largest problems for older adults is the imbalance of participation to conversation. Our approach is to develop a system that assists group conversation of older adults, by evaluating speech duration and facial expression of each participant during conversation. It enables all participants evenly to take part in the conversation, which tends to be difficult for older adults. We analyzed 15 group conversations for calculating the parameters used for evaluating the conversation. The effectiveness of the system using the obtained parameters was validated through the experiment. The system which evaluates only speech duration of each participant and the system which evaluates both speech duration and facial expression of each participant during group conversation were compared by analyzing 8 group conversations for each. The results demonstrated that the system which evaluates both speech duration and facial expression of each participant can make all participants to take part in the conversation evenly and actively. We successfully developed the system that can support group conversation regardless of the contents.

Development of a Sound Source Localization System for Assisting Group Conversation

In this study, we developed a sound source localization system, which consists of Jellyfish-02 and HARK robot audition software, in order to reduce the number of wires for evaluating speech duration. Sound source localization performance of Jellyfish-02 is evaluated by precision, recall, and F-measure. Performance of Jellyfish-02 is superior to conventional microphone arrays. During the experiment, we found that F-measure becomes smaller as the number of speakers increases. We investigated the percentage of speech overlapped periods in natural conversation for the purpose of examining the applicability of the system to measure speech duration in group conversation. From the results, Jelyyfish-02 surpasses conventional microphone array in design and usability. It is potentially applicable for assisting group conversion by measuring duration of speech for each participant.

A Robot Presenting Reproduced Stories among Older Adults in Group Conversation

Group conversation, one form of social activities, plays main roles to train and rehabilitate cognitive function as well as improve emotional states in older adults. It has been mainly utilized for healthy older adults. This paper proposes a robot that warms up group conversations of older adults by reusing or repeating speech statements, which are played successfully to activate group conversations of older adults previously. A novel group conversation technique called the coimagination method for preventing mild cognitive impairments and dementia, was used for collecting and reusing conversation data. Two types of group conversation experiments were conducted among older adults. 1) All participants who were human in coimagination sessions, present their original stories with pictures according to selected topic. 2) One of participants in coimagination sessions was a robot, which presents the reproduced interesting stories. These reproduced stories were collected and implemented into the robot in advance. We analyzed the data by the frequency of evoked laughter in each topic and in all participants. The reproduced stories presented by the robot created more laughter than the original stories presented by human. The robot successfully elicited more laughter than the human participants. Based on these results, we found that the robot successfully enlivened group conversation through evoking laughter.

Development of a Tool for Assisting Group Conversation by Re-Voicing Supportive Responses

Interactive group conversation is very important for improvement of cognitive function and prevention of dementia. Interactive means that all the participants participate equally and actively into group conversation. How to make group conversation more interactive for all participants is a challenging task. Main purpose of this study is to develop the system that assists group conversation, which is capable of dealing with following things, face detection and recognition, speech certain contents recognition, repetition or re-voice of supportive responses based on the result of speech contents recognition, and sound source localization using less sensory environment. We did dialogue experiments with arbitrary pairs among seven young adults and analyzed 14 dialogue data for calculating correct recognition rate of face and speech certain contents, for synthesizing speech based on the result of recognition, and for calculating sound source. The performance of the developed system was preliminary verified through the experimental results. The developed system can be used as a basis of assisting group conversation.

Development of Laughter Motion on the Cognitive Robot Bono-02 Assisting Group Conversation

In recent years, Japan is facing the problem of super-aged society. One problem of super aged society is that the number of patients with dementia is increasing. People who dont have much social interaction have higher risk to occur dementia. We have developed a robot, which can turn take and prompt speech of participants via supportive responses in order to activate group conversation among older people. However, motion for turn taking or prompting speech have not been investigated. Purpose of this study is to investigate appropriate laughter motions of the robot by applying several laughter motions of active human speakers. Experiments were carried out among 14 young people. Then we evaluated the laughter motions of the robot as average smile degree of the participants objective evaluation and as points of questionnaires answers subjective evaluation. The experimental results showed that the sounds fu-fu-fu and ha-ha-ha of laughter of the robot were suitable with the motion of moving ones hand in front of mouth, and the sound wa-ha-ha-ha went well with motions of bending the upper part of the body and extending both hands.

Motion Design-A Gel Robot Approach

The main focus of this chapter is to propose methods for motion design of deformable machines, using a particular electroactive polymer gel. They are deformable like mollusk that can locomote dynamically or manipulate things softly. Such a machine has been a dream in the past but is now experimentally possible. Mechanisms consisting of the gel, hereafter called ‘gel robots’, were designed, developed, and controlled experimentally. It includes: (1) a mathematical model of the gel to be applied for design and control of distributed mechanisms, (2) gel robots driving systems, (3) control of gel robots for dynamic deformations. This chapter overviews a gel robot approach based on agent model for motion design of deformable robots utilizing electroactive polymers with simulation and experimental results.