In 2015, the population of people over the age of 65 in Japan was 25.0%. This means that Japan has already become a super-aging society. As such, the number of elderly people living alone has also increased. For such people, a fall is serious because it can lead to severe injury or death. In this paper, we propose a mobile robot to detect human falls and report them to their observers. The mobile robot consists of a household mobile robot (Yujin Robot's Kobuki), a sensor (Microsoft's Kinect), and a computer (PC) to detect the human and control the robot. Although this robot has a method of detecting a fall while following the elderly in a mobile style, the problem was that the fall detection rate was low. One of the factors is the low recognition performance of the object to be tracked. To solve this problem, improvements to the input accuracy by reviewing the recognition sensor and speed adjustment, obstacle detection / avoidance, and falling detection range must be made. In addition to the improvement of the movement detection and speed adjustment of basic subjects as a result of the evaluation experiments, it was a problem that the follow-up rate drops due to obstacle avoidance. About the contents that improved the fall detection rate by making it possible to adjusting the tracking range and re-evaluating it. As a result of the re-evaluation, it is possible to follow the left and right, front and back of the subject.
ASJC Scopus subject areas
- Computer Science(all)