Demands for interaction designs of therapy robots is projected to increase as the expansion of robot-assisted activity. In consideration of this situation, I compared the influence of users' personality traits on the impression of tolerant, rejective and emotional reaction robots to approach from users. As a result, I confirmed that extraversion and openness personality traits have significant influences on the extroversion and steady in the tolerant reaction robot, and the tolerance in the emotional reaction robot, respectively. I also confirmed that the rejective reaction robot creates significantly the narrow-minded than the others independently of users' personality traits. These results suggest that in designing interactions of therapy robots, it is better to design avoiding rejection reactions for any robot-assisted activity. In addition, it should be noted tolerant and negative reactions in designing interactions of therapy robots for robot-assisted activity expected to have many users who are high in extroversion and openness, respectively.

The asynchronously tuned cellular automaton (AT_ECA) we proposed has been shown to generate critical spatiotemporal patterns without fine-tuning of order parameters. In this study, we propose a learning system that applies the remarkable characteristics of AT_ECA to reservoir computing, which has recently been attracting attention as a learning model for time series data. Then, the learning ability of the proposed system was evaluated by the learning task called five-bit task. As a result, it became clear that the success rate of learning is relatively high even if distractor in time series data to learn is long.

Recently, the application of artificial intelligence for the robotics is expected. On the other hand, both the development of robots and AI need high technological skills. A lot of frameworks or tools are released for developing robots or neural networks easily. For example, in the robot region, OpenRTM-aist has OpenRTP which is an Eclipse based tool for making software of RT Middleware. And also, in neural network region, Keras is one of the easy developing frameworks with TensorFlow and Theano. There are many good frameworks and tools, however, there is a high wall between the regions: tools for robot systems are not for neural networks, and vice versa. It is necessary to build a new environment to develop robot systems and neural networks. So, we have developed an IDE “AirGraph” for the developing of robots with AI system, which can be used easily and simply. In this paper, first, we introduce the idea about the IDE and a case study of it. Then, we suggest the “AirGraph” including the implementations to solve issues found in the case study.

Japan has geographical conditions that make it prone to being affected by natural disasters. In the event of a disaster such as an earthquake or typhoon, floods, landslides and buildings collapse. In this situation, evacuation routes that have been determined may change dynamically. We focused on the possibilities of drone. In this research, we obtain safe guidance route in dynamic environment and guide victims and the headquarters can watch these information by the drone. As a result, using “ROS” (robot operating system) for system implementation and we checked the logic in the simulation.

In this paper, we assume a situation that multiple quadrotors are flying autonomously in the same space for the purpose of aerial photographing, surveillance, home delivery and so on. In such a situation, an algorithm to avoid mutual collision is needed. D. Zhou et al. proposed an algorithm based on the buffered Voronoi cells (BVC) to reach each target position without mutual collision when the multiple quadrotors fly and confirmed its operation by some simulations. In this research, we verified the performance of the method by numerical simulations using MATLAB. In addition, we confirmed the algorithm by dynamic simulations with four quadrotors using the dynamic simulator V-REP.