A High-Precision Vision-Based Mobile Robot Slope Detection Method in Unknown Environment

Aiming at satisfying the demand of slope perception accuracy for mobile robots in the unknown environment, a high-precision vision-based mobile robot slope detection method in unknown environment is proposed in this paper. Firstly, the RGB images of slopesin the unknown environment arecaptured by the Kinect sensor installed on the mobile robot, and edge sharpening is performed on the images of slopes. Then, the coordinates of the slope edge pixel in the image coordinate system are transformed into the camera coordinate system to obtain the real length of the slope. Finally, the single image depth estimation method based on convolution neural network (CNN) is used to estimate the depth of theslope image toobtain the projection length of the slope only using a single slope image, and the slope angle is obtained by the inverse cosine function. The experimental results show that the slope detection method proposed in this paper has higher detection accuracy and can satisfy the demand of slope sensing accuracy of mobile robot in unknown environment.