Ask Your Question
0

Is this a normal latency for ROS with Turtlebot?

asked 2017-07-24 15:07:29 -0500

underactuated gravatar image

I am new to ROS and I know little about what to expect from it and from my Turtlebot2 in terms of performance. It does seem to me that it works too slow, but I do not know for sure. I would like someone to confirm this, and maybe suggest ways to diagnose my setup to find the reason of this slowness.

Everything looks good if nothing on the workstation is subscribed to the camera feeds:

$ rosnode ping --all
Will ping the following nodes: 
 * /zeroconf/zeroconf
 * /camera/driver
 * /camera_base_link
 * /master
 * /capability_server
 * /cmd_vel_mux
 * /camera/depth_metric
 * /robot_state_publisher
 * /bumper2pointcloud
 * /camera/depth_rectify_depth
 * /camera/depth_points
 * /interactions
 * /app_manager
 * /camera/depth_metric_rect
 * /capability_server_nodelet_manager
 * /mobile_base
 * /rosout
 * /camera/rectify_color
 * /camera/camera_nodelet_manager
 * /turtlebot_teleop_keyboard
 * /rviz
 * /camera_base_link1
 * /camera_base_link3
 * /camera_base_link2
 * /diagnostic_aggregator
 * /camera/points_xyzrgb_sw_registered
 * /camera/register_depth_rgb
 * /mobile_base_nodelet_manager

pinging /zeroconf/zeroconf with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:51171/  time=7.439137ms
pinging /camera/driver with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:46037/  time=8.575916ms
pinging /camera_base_link with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:59431/  time=7.348061ms
pinging /master with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:35006/  time=36.192894ms
pinging /capability_server with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:49286/  time=11.957169ms
pinging /cmd_vel_mux with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:35104/  time=20.205975ms
pinging /camera/depth_metric with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:42717/  time=92.966080ms
pinging /robot_state_publisher with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:59314/  time=8.183002ms
pinging /bumper2pointcloud with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:45636/  time=13.126135ms
pinging /camera/depth_rectify_depth with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:53540/  time=7.639170ms
pinging /camera/depth_points with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:43682/  time=11.249065ms
pinging /interactions with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:54360/  time=13.771057ms
pinging /app_manager with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:38159/  time=15.375137ms
pinging /camera/depth_metric_rect with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:38714/  time=10.761976ms
pinging /capability_server_nodelet_manager with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:58260/  time=7.064819ms
pinging /mobile_base with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:47401/  time=25.177002ms
pinging /rosout with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:47568/  time=17.100096ms
pinging /camera/rectify_color with a timeout of 3.0s
xmlrpc reply from http://192.168.xx.x16:60723/  time=17.340899ms
pinging /camera/camera_nodelet_manager with a timeout of 3.0s
xmlrpc reply ...
(more)
edit retag flag offensive close merge delete

Comments

At the bottom of the snippet from rosnode ping --all it says ERROR: Unable to communicate with master!. Are you having network issues?

jayess gravatar image jayess  ( 2017-07-24 17:43:39 -0500 )edit

@jayess, it seems I do have network issues when something subscribes to the camera streams, like rviz or rostopic hz. It often results in significant lags of various ROS commands, like rosnode list. Teleop becomes unresponsive and so on. Is it normal?

underactuated gravatar image underactuated  ( 2017-07-24 20:30:56 -0500 )edit
1

It depends on your setup and environment. From my experience, streaming video over Wi-Fi can have significant lag. Sometimes on the order of 10s of seconds or longer.

jayess gravatar image jayess  ( 2017-07-24 20:33:06 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted
2

answered 2017-07-25 13:04:08 -0500

tfoote gravatar image

From your comments of degrated performance when you start streaming camera data it's likely you're saturating the wireless link with that data and it's preventing other connections. ROS data streams are quite capable of saturating network communications, especially wifi. And if the wifi is saturated, connectivity issues are to be expected.

I'd recommend that you look at using a throttle node to subscribe to a lower data rate topic when subscribing remotely.

edit flag offensive delete link more

Comments

I think you are right. I did just try throttling a topic and testing it with rostopic hz and rosnode ping --all. The ping times do increase, but within a limit; the network does not get clogged. Thanks.

underactuated gravatar image underactuated  ( 2017-07-25 13:50:27 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2017-07-24 15:07:29 -0500

Seen: 270 times

Last updated: Jul 25 '17