Estimating the 2D Static Map Based on Moving Stereo Camera

Perception is an essential procedure for intelligent vehicles where the safety issue is the most critical one. Usually, the perceptual approach is constructed based on measurements received from multiple sensors such as (Radar, and LiDAR) in order to model the immediate driving environment for autonomous vehicles navigation. These sensors are often limited and uncertain in providing visual information in any weather condition and they are expensive. Furthermore, they require intensive calculations. Therefore, they can't be easily processed online. The aim of this study is to provide a solution based on the low-cost, light, and low-power stereo camera. The proposed solution focuses on the spatial information about the driving environment which is represented as a 3D point cloud. These post-processed points are projected on the 2D rectangle grid and divided into identical square cells. Each cell is holding information about the 3D points that lie over it and this created what is called a height map. In the same time, a confidence map is built to reduce and discard scattered points because the produced 3D point cloud is noisy. Finally, an occlusion map is constructed to estimate the status of each cell as a border, free or occluded.