Most AI and robotics software are written in Python and C++. This is a best way to hardware level. In this code you can use deep learning to stereo disparity estimation 
GitHub
GitHub - NVIDIA-ISAAC-ROS/isaac_ros_dnn_stereo_depth: NVIDIA-accelerated, deep learned stereo disparity estimation
NVIDIA-accelerated, deep learned stereo disparity estimation - NVIDIA-ISAAC-ROS/isaac_ros_dnn_stereo_depth



