基于图像识别技术的机器人移动中障碍物定位研究

Research on Obstacle Location in Robot Movement Based on Image Recognition Technology

在进行机器人移动中障碍物定位时,通过直方图反向投影法完成图像信息检测,定位误差大,因此该文提出基于图像识别技术的机器人移动中障碍物定位方法。依据双目立体视觉技术原理构建测量模型,实现机器人运行环境图像的动态采集,通过图像处理、图像识别、图像解码等环节明确环境图像中的障碍物信息,结合小波不变矩特征,识别机器人移动中的障碍物,引入投影差分法通过二值化后的逆投影差分图得到障碍物定位结果。实验结果表明,该文方法的障碍物定位结果误差最大值仅为0.5 cm,相比对比位方法,使得最大定位误差大幅度减少。

When locating obstacles in robot movement,image information is detected by histogram back projection method,and the location error is large.Therefore,an obstacle location method in robot movement based on image recognition technology is proposed.According to the principle of binocular stereo vision technology,the measurement model is constructed to realize the dynamic acquisition of the robot running environment image.The obstacle information in the environment image is defined through image processing,image recognition,image decoding and other links.Combined with the wavelet moment invariant feature,the obstacles in the robot moving are identified.The projection difference method is introduced to obtain the obstacle location results through the binary inverse projection difference map.The experimental results show that the maximum error of the obstacle location result of this method is only 0.5 cm,which greatly reduces the maximum location error compared with the comparison position method.