基于单目相机与里程计的室内机器人定位研究

昂贵的定位传感器限制了室内服务机器人相关技术在实际生活中的发展。室内服务机器人定位技术的发展趋势之一是使用价格低廉的传感器。因此,基于单目相机与轮式里程计两种价格低廉的传感器,提出一种适用于室内动态环境的机器人定位解决方案。该方案将相机朝向天花板,以减少室内动态环境下信息多变带来的误差,改变特征匹配策略以提升重复纹理场景下的匹配精度,并结合轮式里程计固定场景尺度。在典型室内场景进行验证的结果表明,在仅使用天花板信息进行定位时,该方案具有较高的定位精度。

Research on Indoor Robot Localization Based on Monocular Camera and Wheel Odometry

The expensive positioning sensor limits the development of indoor service robot related technology in real life.One of the development trends of indoor service robot positioning technology is the use of low-cost sensors.Based on the two low-cost sensors,monocular camera and wheel odometer,a robot positioning solution is proposed which is suitable for indoor dynamic environment.The scheme will reduce the error caused by the changeable information in the indoor dynamic environment by facing the ceiling,change the feature matching strategy to improve the matching precision in the repeated texture scene,and fix the scene scale with wheel odometer.The results of the verification in typical indoor scenes show that the scheme has high positioning accuracy when only ceiling information is used for positioning.