Road Slope Estimation Using Field-Adjustable Stereo Vision

Modern vehicles rely on springs or hydraulic pistons to absorb road irregularities. Accurately estimating road surface height can greatly enhance a robot’s navigation ability. This paper proposes an improved camera setup, using an adjustable-angle camera to capture more comprehensive road surface data, with real-time correction reducing errors to $1-2 \%$ and maintaining 60 FPS. The system converts 2 D images into 3D point cloud data and introduces a method for calculating translation matrix parameters, improving data correction accuracy. Results show significant improvements in detecting slopes, especially downhill, and the system adapts to terrain changes, enhancing the safety of autonomous driving.