if your camera and vision sensor are at the same pose, same view angle and same viewport proportions, then you should see the robot in a similar way left and right. Keep however following in mind:
a camera will display objects that are in the current view layers
a vision sensor will display objects that are renderable, independent of the view layers they are in
when looking at a model via a camera, you may only see the currently visible parts... other parts may be hidden in a layer that is currently not shown. This is often used with robot models, where the visible parts are detailed and used for collision detection, distance calculation and proximity sensor simulation, and where the hidden parts are simplified and often used for dynamics. If your visible parts and hidden parts do not overlap, then you have a problem, and this is also what might be going on in your scene. Try to look at your scene by toggling the visible layers a few times (also during simulation). You will quickly see if something is not right.
Thank you for your answer.
All objects are set to be visible in all layers and to be renderable. I'm not using different structure parts for a different purpose.
The problem occurs only if joints are not strong enough. When I changed max torque on joints to higher value then this weird behavior disappear.
Any idea how to solve it? I can't use higher values for joints torque.
coppelia wrote: ↑30 Jan 2020, 06:19
Can you post a simple scene where that strange behaviour occurs? It will be easier for us to check what is going on...
Cheers
Hello,
Sorry for my late reply.
Cheers
Last edited by zdenoB on 30 Mar 2020, 19:16, edited 1 time in total.
The reason why your vision sensor and camera are not showing the same behaviour is because you perform some actuation in the sensing section of your child script. So what is happening in each simulation step:
1. regular actuation
2. vision sensor is taking an image (in the sensing section)
3. you are changing things in the scene (in the sensing section)
4. the camera is taking an image
In the sensing section you should normally only do sensing tasks.