基于QNX的双足机器人视觉姿态平衡研究

由于仿人双足机器人的类人运动特点,需要采用平衡控制方法实现其姿态平衡控制。目前平衡控制方法多采用外界传感器反馈的形式获取姿态信息,系统结构复杂。提出一种基于机器视觉的双足机器人姿态平衡控制方法。以韩伯机器人为机器人原型,采用嵌入式系统开源可扩展硬件平台为机器人控制板,在QNX嵌入式实时操作系统上实现机器人直立状态下的姿态平衡控制。首先,控制板获取机器人头部摄像头的图像数据,对图像数据进行二值化处理。然后,按照视觉姿态倾角模型计算出机身倾斜角度,并通过串口向动作控制板发出姿态调整命令。最后,由动作控制板完成姿态调整动作。实验结果是动作控制板按照调整命令,控制各数字伺服马达工作,在误差范围内实现机器人直立状态下的姿态平衡控制。

The Research on Bipedal Robot Posture Balance Controlling Method with Robot-vision Based on QNX

Because of the movement characteristics of humanoid bipedal robot,it needs controlling methods to make the robot balanced.Most of the existing methods use sensors to feedback the posture information,and the system architecture is complicate.This paper presents a robot posture balance method based on machine vision.In this paper,the system uses HBE robot as the prototype,the embedded system scalable open source hardware platform as the robot control panel and realizes the robot control to achieve a balanced posture on QNX operating system.Firstly,using the main panel to capture image from the camera on the robot head and then binary the image.Secondly calculating the tilt angle by using the vision-balancing model and sending command to the movement controlling panel.Lastly,making the robot to be balance again.Experimental result is that the movement controlling panel makes the servos work well according to the angel commands and achieves the balance controlling method within the error range.