Skip to content

Latest commit

 

History

History
64 lines (32 loc) · 1.51 KB

File metadata and controls

64 lines (32 loc) · 1.51 KB

Free Space and obstacle detection using occupancy grids

Tools: OpenCV 2.4, C++, CUDA.

This impelementaion is based on Free Space Computation Using Stochastic Occupancy Grids and Dynamic Programming with some modifications.

The code is tested with Zed stereo camera and nvidia jetson tk1, and the parameters depend on the camera(calibraion) and the position of the camera.

This repo is still under development.

To make cpu and gpu code

type make then make run

To make CPU only, remove #define GPU from main.cpp

type make cpu then make run_cpu

To make GPU only, remove #define CPU from main.cpp

type make gpu then make run_gpu

Input image and its Depth

Polar Occypancy Grid with different sizes

Cartesian Occypancy Grid with different sizes

Segmentation of polar grid

Thresholding

Free space

Dynamic programming

Free space

GPU Implementation