Strange data when using force sonsor

Report crashes, strange behaviour, or apparent bugs
Post Reply
kou5321
Posts: 2
Joined: 30 Mar 2021, 13:59

Strange data when using force sonsor

Post by kou5321 »

my CoppeliaSim version is EDU 4.0.0. And I want the robots to use line-following the trajectory to push some kinds of objects. But I found that even I set the objects in the same parameter, sensor data will change as the position and size of objects. Some of the data is correct while some data is shaking and very strange.

the position of the force sensor is between a cylinder and robot link. Pictures could be seen in:
https://sm.ms/image/CednHcE6Lat2hxI

correct push picture seems to be a straight line as the ink followed:
https://sm.ms/image/BIjoMWRYxHaFnmz

incorrect data either cannot get data in the process for some time(as you see, the process is going on, but the data has stopped for a long time):
https://sm.ms/image/vXF1zAETjGeftJP

and sometimes the data can just shake violently:
https://sm.ms/image/wq3CEaDGLpV7FeJ

The attributes of the objects are set the same, including masks, dynamic properties. The whole file can be seen in
https://drive.google.com/file/d/1IOVwdy ... sp=sharing

Many thanks in advance.

coppelia
Site Admin
Posts: 8974
Joined: 14 Dec 2012, 00:25

Re: Strange data when using force sonsor

Post by coppelia »

Hello,

I don't completely understand your question. But from what I can see (the image links are extremely slow, and some fail to load the image), it is not something exceptional when the sensor reading is shaking. Make sure to try with another physics engine too, and adjust also the dynamics configuration mode to see if data improves (i.e. from accurate to very accurate, etc.).

Also, you won't get the same readings across physics engines, since friction is implemented in different ways depending on the engine.

Also, make sure that the cylinder that connects to the force/torque sensor is not colliding with the arm itself (you should set its local respondable mask to zero).

Cheers

kou5321
Posts: 2
Joined: 30 Mar 2021, 13:59

Re: Strange data when using force sonsor

Post by kou5321 »

Thanks for your reply. It really helps me for the work! I still have two questions for this.

It seems that different physical engine just produces quite different data in doing the same action. I made several experiment for this. The result you can see in link:
https://drive.google.com/drive/folders/ ... sp=sharing

It seems that the data produced by each physical engine varys a lot. And I just wondering whether I can get accurate and stable data in one engine? (As you know, in some cases when different engines produce different data, it is hard to decide which one to use)

Another question is that when I use Vortex which I think it will be accurate, the object just fall directly... The video is also added in the link above. Would you offer an idea how to set objects' propoties so I can use Vortex properly.

Thanks a lot.

coppelia
Site Admin
Posts: 8974
Joined: 14 Dec 2012, 00:25

Re: Strange data when using force sonsor

Post by coppelia »

This is the nature of a physics engine, and you will have to look at the various parameters and settings for each engine, in order to produce results that satisfy you. Usually more calculation iterations brings more stability and precision, but it really also depends on the application.

As for the Vortex engine, you will have to download Vortex Studio and register for a key with CMLabs. Only then will the Vortex engine be made available/operational.

Cheers

Post Reply