Odometry and compass odometry pose estimate comparison

Post links to your CoppeliaSim videos, scenes and models
Post Reply
Eric
Posts: 186
Joined: 11 Feb 2013, 16:39

Odometry and compass odometry pose estimate comparison

Post by Eric »

http://youtu.be/wLBwvDqBcuA?list=UU-OW0 ... 7pQNnoK4Lw

This video illustrates two dead reckoning pose estimations.

The blue robot is a differential drive robot (i.e. the ground truth). It is dynamically simulated so that wheels can slip.

The Red robot is a representation of the pose estimate based on pure odometry (data gathered from wheels rotation via "virtual" encoders.

The green robot is a representation of the pose estimate based on compass odometry. Compass odometry is using the perfect orientation reference instead of the estimate given by the wheels' encoders.

Balakrishnan1992
Posts: 11
Joined: 04 Apr 2015, 14:17

Re: Odometry and compass odometry pose estimate comparison

Post by Balakrishnan1992 »

I wish to find the orientation of a differential drive robot. Can I do the same with a compass in vrep. Where is the compass located ?(which library I mean)

coppelia
Site Admin
Posts: 10312
Joined: 14 Dec 2012, 00:25

Re: Odometry and compass odometry pose estimate comparison

Post by coppelia »

Simply check the orientation of a given object with simGetObjectOrientation.

Cheers

Balakrishnan1992
Posts: 11
Joined: 04 Apr 2015, 14:17

Re: Odometry and compass odometry pose estimate comparison

Post by Balakrishnan1992 »

The thing is I am not allowed to use simgetobjectorientation on my robot. Is there another way to get the orientation without this ? I am afraid to wheel rotation to compute the heading because I am afraid about slippage. Can you suggest me a way to do this ??

coppelia
Site Admin
Posts: 10312
Joined: 14 Dec 2012, 00:25

Re: Odometry and compass odometry pose estimate comparison

Post by coppelia »

If you need the exact orientation, then use simGetObjectOrientation. If you need a measured orientation like on the real robot, you will have to do odometry, and/or use some other sensors to extract an orientation.

Cheers

Post Reply