tf lookuptransform time lag problem
I am trying to use look up transform and tf broadcaster to log the robot position at the certain time. When I run the core for more than 20 mins, lookuptransform cannot receive a tf anymore. And what I observed is that lookuptransform actually lags what shows up in the rviz. I used rospy.Time() for lookuptransform, and rospy.Time().now for sendtransform. Is there any reason for this?
Could you please provide more details?
Are you trying to lookup a transform from 20 minutes ago, or are you saying a TF listener appears to stop receiving new transforms after 20 minutes?
I'm not saying what you observe could not be true, but seeing as TF (and TF2) have been in continuous use for the past 10+ years and the severity of what you describe (it stops working after 20 minutes) I would have expected many more reports about this.
At this point I would be inclined to suspect something local to your setup, a configuration issue or something else not working as it should.
But we'll need more information to say something more useful.
I totally agree with you on it shouldn't be an tf problem. Is there anything that I should look out for? For example, I do a rospy.sleep(1.5) every time after I setup a transform, and could that mess up the time somehow? Should I use time.sleep instead?
Please answer my question and describe with some more detail what you observe:
those are two different things and your current question text could be interpreted to describe both of these.