基于多传感器信息融合的仿人机器人跌倒检测及控制

发布时间：2018-05-26 19:43

本文选题：仿人机器人 + 控制　；参考：《华南理工大学学报(自然科学版)》2017年01期

【摘要】：仿人机器人在行走过程中,需要及时检测到是否要摔倒,从而阻止摔倒或通过保护动作在摔倒过程中减少机器人损伤.文中基于姿态传感器(包括加速度传感器和陀螺仪)、力感应传感器(FSR)及步行阶段建立了多传感器信息融合模型,采用模糊逻辑决策方法建立了机器人跌倒检测的综合判定方法,并设计相应的控制器,该控制器可检测到仿人机器人运动时是否会摔倒,若无法避免摔倒,则采取相应的保护动作;若可避免摔倒,则通过控制髋关节阻止机器人摔倒.最后在SCUT-I仿人机器人上进行实验,结果表明,文中方法能及时检测到机器人在运动过程中发生跌倒的时刻,并及时通过稳定控制器阻止摔倒或通过倒地控制器产生保护动作,减少机器人的损伤.
[Abstract]:In the course of walking, humanoid robot needs to detect whether to fall in time, so as to prevent the fall or to reduce the robot damage by protecting the robot from falling down. A multi-sensor information fusion model is established based on attitude sensors (including accelerometers and gyroscopes, force sensor FSRs) and walking stage. The fuzzy logic decision method is used to establish a comprehensive decision method for robot fall detection. A corresponding controller is designed, which can detect whether the humanoid robot will fall while it is moving, if it cannot avoid falling, it can take corresponding protective action; if it can avoid falling, it can prevent the robot from falling down by controlling the hip joint. Finally, experiments are carried out on SCUT-I humanoid robot. The results show that the method in this paper can detect the time of the robot falling in the course of motion in time, and stop the fall by stabilizing the controller or produce the protection action by the falling controller in time. Reduce robot damage.
【作者单位】：华南理工大学计算机科学与工程学院;华南理工大学软件学院;
【基金】：广东省-教育部产学研合作专项资金项目(2013B090500093) 广东省科技计划项目(2015A020219001) 国家自然科学基金面上项目(61372140) 广州市机器人软件及复杂信息处理重点实验室(15180007) 广州市科技创新委员会项目(201609010075)~~
【分类号】：TP242.6;TP212

【相似文献】