Realization of Adaptive gait patterns for a Quadruped Robot with 3-Dimerisional Recognition Ability

具有三维识别能力的四足机器人自适应步态模式的实现

基本信息

批准号：
15560220
负责人：
KAMANO Takuya
金额：
$ 2.37万
依托单位：
The University of Tokushima
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
2003
资助国家：
日本
起止时间：
2003 至 2004
项目状态：
已结题

项目摘要

Main results of this research are follows :1 Generation of fundamental gait patterns for a quadruped robot with CPG networkIn this research, a new Central Pattern Generator(CPG) within the dynamics of joint drive motor is proposed. The CPGs can drive the joints of leg periodically. The CPGs are connected mutually by the coupling parameters. The internal parameters and coupling parameters are adjusted by genetic algorithm so that the quadruped robot realizes the adequate actions. As a result of tuning, the fundamental gait patterns for three motion, linear motion, right and left turn motion are generated.2 Target tracking action using the tuned fundamental gait patterns based on image information.In this research, the tuned fundamental gait patterns are switched by the image information from small CCD camera mounted on the robot. After the robot searches the target, the robot can approach the target in combination with tuned three fundamental gait patterns.3 3-dimensional avoidance motion for obstacles based on environment recognition abilityTo detect the obstacles in forward direction, the ultrasonic sensors and a photo sensor are mounted on top and bottom of the robot. The detected obstacles are classified into three types of obstacles ; roundabout avoidance obstacles, going under and going over obstacles. To achieve the adequate avoidance, the suitable gait patterns are selected for the detected obstacles.4 Stability control of robot attitude using the upsetting danger degreeThe upsetting danger degree for the quadruped robot is defined. The stability control of the robot is realized on the rocking floor so that the upsetting danger degree reduces to zero. As a result of stability control, the robot body can be kept to flat level.

本研究的主要成果如下： 1 利用 CPG 网络生成四足机器人的基本步态模式在本研究中，提出了一种新的关节驱动电机动力学中的中央模式生成器（CPG）。 CPG可以周期性地驱动腿部关节。 CPG 通过耦合参数相互连接。通过遗传算法调整内部参数和耦合参数，使四足机器人实现适当的动作。通过调整，生成了三种运动的基本步态模式：直线运动、右转和左转运动。2 使用基于图像信息的调整后的基本步态模式进行目标跟踪动作。在本研究中，调整后的基本步态模式为通过安装在机器人上的小型 CCD 摄像机的图像信息进行切换。机器人搜索到目标后，结合调整后的三种基本步态模式，可以接近目标。 3 基于环境识别能力的三维避障运动为了检测前方的障碍物，采用超声波传感器和光电传感器。安装在机器人的顶部和底部。检测到的障碍物分为三类；迂回避开障碍物、冲入和越过障碍物。为了实现充分的躲避，针对检测到的障碍物选择合适的步态模式。4 利用翻转危险度对机器人姿态进行稳定控制定义了四足机器人的翻转危险度。在摇摆地板上实现机器人的稳定性控制，使翻转危险度降至零。通过稳定性控制，机器人本体可以保持水平。