Conversation
src/igvc_cv/igvc_cv/cv_node.py
Outdated
| def image_callback(self, msg): | ||
| frame = self.bridge.imgmsg_to_cv2(msg, 'bgr8') | ||
| gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) | ||
| blur = cv2.GaussianBlur(gray, (11, 11), 33, 33) |
There was a problem hiding this comment.
The arguments of the GaussianBlur function (as well as the Canny and HoughLinesProbability) will be adjusted in the future based on our actual camera image and testing. Would it be better to have them as arguments in this function or will they be manually edited in the code?
Glesko22
left a comment
Glesko22
left a comment
There was a problem hiding this comment.
Look at comments, mostly formatting issues
src/igvc_cv/igvc_cv/cv_node.py
Outdated
| self.declare_parameter("hough_min_line_length", 4) | ||
| self.declare_parameter("hough_max_line_gap", 5) | ||
|
|
||
| # Gaussian Pyramid Up-Sample |
src/igvc_cv/igvc_cv/cv_node.py
Outdated
| # Gaussian Pyramid Up-Sample | ||
| self.declare_parameter("layers", 4) | ||
| self.declare_parameter("pyrInput",[]) | ||
| self.declare_parameter("pyrOutput",[]) |
There was a problem hiding this comment.
I believe line 39 should have a comma instead of a period
src/igvc_cv/igvc_cv/cv_node.py
Outdated
| self.declare_parameter("hough_max_line_gap", 5) | ||
|
|
||
| # Gaussian Pyramid Up-Sample | ||
| self.declare_parameter("layers", 4) |
There was a problem hiding this comment.
Small, but I believe there should be a space between the comma and brackets:
self.declare_parameter("pyrInput", [])
self.declare_parameter("pyrOutput", [])
| white_u = white_u[valid] | ||
| white_v = white_v[valid] | ||
| distances = distances[valid] | ||
|
|
There was a problem hiding this comment.
What does 'distances' do? Right now it looks like its calculated and then never used? Not an issue, just wondering the intention
There was a problem hiding this comment.
it seems that this is the initial step of projecting the detected lines into 3d space, since we were not completely sure in what format to output this information. it is a different question as to whether this distances array can be smoothly integrated into the costmap.
There was a problem hiding this comment.
Are we certain this is the way forward? Has anyone looked into getting distances smoothly integrated into the costmap?
|
On costmap integration: obstacle_layer and voxel_layer accept PointCloud2 topics as input. This node now returns a point cloud, but it must still be integrated into the costmap |
4 participants
Added CV ROS2 Node