On the Optimisation of Intelligent Robots for Error Detection of Shadow Regions

With the continuous development of robotics, the improvement of visual obstacle avoidance has become an important direction to achieve autonomous navigation. However, the presence of shadows seriously affects the robot's visual recognition system, resulting in poor obstacle avoidance. Therefore, this paper considers it necessary to give appropriate optimization for some recognition problems of shadow removal. In such a background, this paper designs a new shadow removal method based on the properties of image colour, pixel luminance value, and chromaticity of the shadow part. The method combines morphological processing and median filtering algorithm for noise removal, and simulation experiments are carried out in a Matlab environment to achieve more satisfactory results. The experiments show that the proposed method has higher accuracy and wider applicability in removing marker shadows, and can effectively improve the visual obstacle avoidance ability of the robot. However, the method is not effective for dynamic image processing, and it is expected to solve this problem by training artificial intelligence models in the future.