/vision_lane_detection

这是一个视觉车道线的，稳定版

Primary LanguageC++

Usage

Get the Code and Compile

mkdir catkin_ws

git clone git@192.168.3.54:root/vision_lane_detection.git

cd vision_lane_detection
git checkout dev

cd ..

git clone git@192.168.3.54:root/trunktech_msgs.git

cd catkin_ws
mkdir src
ln -s $PWD/../vision_lane_detection src/
ln -s $PWD/../trunktech_msgs/hadmap_msgs src/

catkin_make
. devel/setup.bash

Configurations

  • config/gantry.yaml
  • config/proto.txt

config/gantry.yaml

num_pixels_per_meter_x_dir: 30   # number of pixels per meter in the horizontal direction in the IPM image
num_pixels_per_meter_y_dir: 20   # number of pixel per meter in the vertical direction in the IPM image
  • Change the camera topic
camera_topic: /pylon_camera_node/image_raw/compressed
  • refer to the comment in config/gantry.yaml for other parameters

config/proto.txt

  • use ipm_calibration tool to obtain the following 9 parameters (image_to_ipm_tf)
ipm_param {
  width: 448
  height: 448

use_precomputed_H: true

image_to_ipm_tf: -0.23864383
image_to_ipm_tf: -0.59772766
image_to_ipm_tf: 412.28256
image_to_ipm_tf: -0.044471879
image_to_ipm_tf: -1.4135034
image_to_ipm_tf: 689.36011
image_to_ipm_tf: -8.53082e-05
image_to_ipm_tf: -0.0027097312
image_to_ipm_tf: 1
}
  • read the comments in proto/lane_line.proto for usages of other parameters. In general, the default parameters are fine.

Run

roslaunch lane_line gantry.launch

Subscribe to the image topic /lane_detection/visualization to view the detection results.