基于视觉导航技术人形机器人的行走和爬行研究

Research Vision-guided Walking and Climbing by Humanoid Robot

主要研究面向人形机器人行走和爬行的立体视觉问题。障碍检测和物体识别广泛采用图像处理和物体识别技术,数据拟合技术也应用于人形机器人的路径规划。通过实时合成高精度的物体特征检测方法能够提供精准的视觉反馈。提出了基于这些方法和机器人导航系统的立体视觉系统,并且利用人形机器人进行了仿真实验。

This paper presents key ideas of stereo vision based on humanoid robot walking and climbing. Image processing techniques and pattern recognition techniques are employed for obstacle detection and object recognition, data fitting technique is also used to plan path of humanoid robot. High precision visual feedback is provided by combination of real time high precision object feature detection method. The proposed stereo vision based approaches and robot guidance system were evaluated partly by experiments with humanoid robot.