成果報告書詳細
管理番号20160000000548
タイトル*平成27年度中間年報 次世代ロボット中核技術開発 革新的ロボット要素技術分野 人ごみをぶつかりながら安全かつ不快感を与えずに移動する自律移動技術の研究開発
公開日2016/7/22
報告書年度2015 - 2015
委託先名パナソニック株式会社 学校法人早稲田大学
プロジェクト番号P15009
部署名ロボット・AI部
和文要約
英文要約1. Research Purpose It is important for co-robotics to complete the task under the environment where there are many people. Conventionally, the robot is not allowed to contact with people. In this research project, based on the concept that the robot is allowed to contact with people, an innovative autonomous moving technology should be developed to realize co-existence of human and robot in the crowd. Panasonic aims to develop the mechanism for the safety contact, and the autonomous moving algorithm of the robot in case of some physical interactions between human and robot. In Waseda University, we assume typical scenes where a robot passes through near humans in a crowded space. We clarify a movement mechanism of humans during passing through on the basis of results of observing behaviors of humans in the above crowded environments, and analyze basic parameters for constructing a movement-estimation module. Moreover, we analyze not only human behaviors toward robot behaviors, but also reactive human-behaviors toward robot actions.
2. Results
-- Panasonic Corporation
An autonomous moving robot has been developed to complete the design of hardware and software in FY2015.The robot has two driving wheels and one training wheel. A laser range sensor is arranged at 400mm height to detect the wall under the surrounding environment. Three bumper sensors which are attached at the front side of the robot are used to detect objects. The robot has Iterative Closest Point(ICP)as the basic autonomous moving algorithm, which matches the pre-registered map information and the laser range sensor information. By using this method, the robot can travel any positions on the map. When the laser sensor detects some objects which are not registered on the map, the robot judges them as obstacles and avoids the collision.The objects which are not able to be detected by laser range sensor are sensed by three bumper sensors attached on the front side of the robot, and avoided the severe collision. The mechanism and the control algorithm of the developed robot are evaluated. Firstly, the experimental and artificial environment is used to show the effectiveness of the developed autonomous moving robot system. Then, the robot is evaluated at the surrounding environment. It is confirmed that the robot is able to move along the wall and keep moving with contacting with objects under the specific environment.
-- Waseda University
We then conducted experiments, which has various parameters such as distance between human and robot, position of contact, and contact force vector. We observed and analyzed reactive behaviors toward robot motions. The results of analysis indicated that human moves to a direction corresponding to the direction of force vectors applied, on the other hand, these reactive motion nonlinearly changed depending on contact parts of human.
In order to develop a method to convey human-intention which makes both human and robot pass through each other safely, we analyzed ways of conveying intention, referring collaborative movement for human-human cases. On the basis of the analysis, we developed a motion control module which measured relative distance between human and robot, width of corridor, direction of human by using laser range finders, and then selected inductive motions depending on the human reactive model we have already proposed. Here, we implemented a method to determine a pass of robot by using RRT (Rapidly-Exploring Random Trees) during contact with human, and evaluated human’s subjective feelings. Human symbiotic robots should behave naturally in actual environments, like humans. As one way to realize this, we made a prototype of intention conveying methods using whole-body gesture in a proximal space. The experimental results indicated that relative distance between human and robot and movement space largely affects the degree of collaborative movement.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る