Available perception sensors (lidar/radar) are costlier to implement in low/mid-range cars in India/Asia. Cameras are not suitable for perception in all-weather driving. Hence, a low-cost, compact hybrid perception system (HPS) suitable for these markets is under development by combining an indigenously developed sparse lidar (four-channels) with a monocular camera. Data fusion is used to scan and identify obstacles in the environment. On-road tests conducted with a scaled-down vehicle show that HPS can carry vehicle localization, lane centering and emergency braking, and also detect obstacles, faded lane markings, road signs, etc. in highly deteriorated roads.

• Homography estimation for precise transformation of lidar-camera data fusion
• Modified spatial calibration for velocity estimation with monocular camera
• Dynamic thresholding adaptive road region estimation for lane detection
• Robust Vehicle pose estimation using HPS and ground spatial