Closed Juhyung-L closed 11 months ago
Hey @Juhyung-L Did you find any solution to this? I am facing a similar issue. Although I can see the image in Rviz, it is very blurry, I have tried increasing the image resolution as well as the env_texture_size, but no changes.
Hi, sorry for the late response.
My problem was not blurry images. It was that I was using 3 wideanglecameras
in Gazebo, which was making the simulation very slow. I still have not found out why using wideanglecameras
is slow. So I just decided to use the regular camera
.
OS: Ubuntu 22.04 Gazebo Version: 11
I have 3 cameras attached to a box (right, left, and back sides of the box).
When the camera type is
type="camera"
, the real time factor is around 0.6-0.7 When I change the type totype="wideanglecamera"
, the real time factor drops to ~0.1I get that
wideanglecamera
captures a wider view and is more computationally expensive to simulate, but even when I set the FOV of thewideanglecamera
to be the same ascamera
,(I set it to 90degrees/1.5708radians) the RTF is still ~0.1.urdf for wideanglecamera
urdf for camera
I am using a ROS2 launch file and I get some error messages when I spawn the robot model with the
wideanglecamera
robot_world.launch.py
spawn_robot.launch.py (used by robot_world.launch.py)
The error messages in terminal output:
I did some search on the error message, but could not find anything useful.
So my two questions are:
wideanglecamera
so expensive to run even if I set the FOV to be 90degrees whencamera
with the same FOV is not as expensiveNode::Advertise(): Error advertising topic [...]. Did you forget to start the discovery service?
I am running ROS2 and Gazebo from a Docker container (I shared the host machine's XServer with the container for GUI)
This is how I installed Gazebo in the contianer