Invited Talks

Abstract:

Now many business fields are facing the aging issue and shortage of the labor. We produce many kinds of Powered Wear aim to achieve the society where people can live without being affected by age or gender. Powered wear brings out the potential of its wearer in near future it will become similar to a shirt or a pair of pants. When the robots are wearable, the human will get more physical freedom.

Abstract:

We have been developing the wearable muscle suit(R) (Exo Muscle(R)) directly support motion. Use of the McKibben artificial muscle makes muscle suits lightweight and practical. We have started selling “Lower Back Assist” type since Nov., 2013 and more than 4,000 units have been distributed in Japan for manual workers in many kinds of field such as logistics, care giver, agriculture, building industry, and so on. Also we started selling the arm and lower back support muscle suit since Sep. 2018. Both of them are possible to use in active and passive way. Mechanism and future plan will be presented.

Abstract:

We develop a farm management system that uses images from drones and Global Positioning System (GPS) sensors attached on the grazing cows. Based on a lot of photos taken by a drone, the calculation of the ratio of weeds in each farm area helps in deciding the date to remove the broad-leaved weeds instead of having the farmer walk around the entire farm. Using Internet of Things (IoT) data about cows and the grass, the farmer can figure out the places that need the fertilizers. The proposed region segmentation based on the deep-learning method could detect the broad-leaved weeds with an accuracy of around 80 %. The IoT data-based suggestion could calculate the place which has no grass, smooth gradient, and fewer cows.

Abstract:

Natural interaction is important for both Human Computer Interaction (HCI) and Human Robot Interaction (HRI). The purpose of this topic is to enrich and reimage the human computer or robot interfaces. We take a nursing-care robot as an example to investigate the idea of natural interaction. This study is an attempt to explore the adaptive human robot interaction and contributes to giving insights and implications for the future design of general serving robot. I will introduce how contextual information such as the relationship between the hand, the pen, and the tablet can be leveraged in the digital drawing experience to further enhance its naturalness. I will also show some demos and project ideas of our recently projects.

Abstract:

In our research, we focus on how to measure or estimate the human motion and the external force to calculate the burden of user when supported by a power assist device. In order to make the measurement system easy-to-use, a method was proposed to estimate the human motion only using a few IMU-based motion sensors. We also proposed a method to estimate the external force (the ground reaction force and the operating force) only from the measured human motion. In order to control the device without delay, we also proposed a method to predict the motion state from the plantar forces or some IMU sensors based on the machine learning technology. In this talk, I will introduce these methods and show you the results of the experiments.

Abstract:

Recently, soft robots have received tremendous interests for their potential applications. As one kind of soft actuation materials, dielectric elastomer actuator (DEA) has been paid more attention because of its prominent human-muscle like features which are light, low stiffness, large stretch rate, enough power and acceptable response time. In order to make DEA deform on uniaxial direction as human-muscle, tube-type DEA is developed. However, DEA has hysteretic behavior that may significantly degrade DEA’s performance. For compensating the effect of the hysteresis, a modified play-type Prandtl-Ishlinskii hysteresis model is built to describe the tube-type DEA. Then, based on properties between play and stop hysteresis operators, a parallel compensator is constructed by using stop-type Prandtl-Ishlinskii model. For rejecting the effect of input disturbance and plant perturbation, disturbance observer (DOB) is designed. Finally, the effectiveness of the proposed method is verified by experimental results.