Wezzoid's profile - activity

I am successfully using various types of visualization markers, but can't get TEXT_VIEW_FACING to work. RViz is subscribed to the topic, and the message content seems ok. RViz isn't showing any errors. Am I missing something simple or is there a problem with RViz? Using Kinetic on Ubuntu 16.

Here is the output of rostopic echo. (There is one text marker each for 4 legs, only one shown)

---
header:
  seq: 13963
  stamp:
    secs: 1541615488
    nsecs: 232728531
  frame_id: "/base_link"
ns: "markers_footforce_text"
id: 3
type: 9
action: 0
pose:
  position:
    x: 0.14822583112
    y: -0.163636099005
    z: -0.183330821392
  orientation:
    x: 0.0
    y: 0.0
    z: 0.0
    w: 1.0
scale:
  x: 0.05
  y: 0.05
  z: 0.05
color:
  r: 1.0
  g: 0.0
  b: 0.0
  a: 1.0
lifetime:
  secs: 0
  nsecs:         0
frame_locked: False
points: []
colors: []
text: "hello"
mesh_resource: ''
mesh_use_embedded_materials: False
---

update: Here is an example of a marker msg that works:

---
header:
  seq: 7019
  stamp:
    secs: 1541670264
    nsecs: 695540526
  frame_id: "/base_link"
ns: "markers_footforce"
id: 3
type: 0
action: 0
pose:
  position:
    x: 0.137246076547
    y: -0.150541167427
    z: -0.0882269625985
  orientation:
    x: 0.0
    y: -1.0
    z: 0.0
    w: 1.0
scale:
  x: 0.0925518925519
  y: 0.02
  z: 0.02
color:
  r: 0.0
  g: 1.0
  b: 0.0
  a: 1.0
lifetime:
  secs: 0
  nsecs:         0
frame_locked: False
points: []
colors: []
text: ''
mesh_resource: ''
mesh_use_embedded_materials: False
---

update: Here are my RViz startup messages:

[ INFO] [1541671551.052185041]: rviz version 1.12.16
[ INFO] [1541671551.052267087]: compiled against Qt version 5.5.1
[ INFO] [1541671551.052279569]: compiled against OGRE version 1.9.0 (Ghadamon)
[ INFO] [1541671551.746811300]: Stereo is NOT SUPPORTED
[ INFO] [1541671551.747030266]: OpenGl version: 3 (GLSL 1.3).

Last few lines from rviz -l :

[ INFO] [1541672551.298442008]: GLXWindow::create used FBConfigID = 163
[ INFO] [1541672552.074872851]: Mesh: Loading rviz_sphere.mesh.
[ INFO] [1541672552.080553706]: Texture: SelectionRect0Texture: Loading 1 faces(PF_R8G8B8A8,1x1x1) Internal format is PF_A8R8G8B8,1x1x1.
[ INFO] [1541672552.083017394]: Mesh: Loading rviz_cylinder.mesh.
[ INFO] [1541672552.083373321]: Mesh: Loading rviz_cone.mesh.
[ INFO] [1541672565.843360810]: Mesh: Loading rviz_cube.mesh.
[ INFO] [1541672567.908020785]: Font Liberation Sans using texture size 256x128
[ INFO] [1541672567.911394493]: Info: Freetype returned null for character 160 in font Liberation Sans
[ INFO] [1541672567.912111099]: Texture: Liberation SansTexture: Loading 1 faces(PF_BYTE_LA,256x128x1) Internal format is PF_BYTE_LA,256x128x1.

It's saying it couldn't find a texture from the font for the nbspace char, not that it's actually broken, right?

Update: The interactive markers tutorial demo doesn't show any text either, but works otherwise. Must be something wrong with RViz.

Update re Ogre and fonts:

wes@ubuntu:~$ sudo dpkg -l *ogre*
[sudo] password for wes:
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name           Version      Architecture Description
+++-==============-============-============-=================================
un  libogre-1.8-de <none>       <none>       (no description available)
un  libogre-1.8.0  <none>       <none>       (no description available)
ii  libogre-1.9-de 1.9.0+dfsg1- i386         3D Object-Oriented Graphics ...

I'm trying to make a stereo camera for visual odometry. My Python script to publish the images consistently stops after ~500 images, with no errors.

I have cameras plugged into the 2 CSI ports on the standard Raspberry Pi Compute Module breakout board. I'm using Kinetic on Raspian. I'm using the picamera library so I can access the cameras as if I was using raspivid. This brings in both cameras as one image with good sync (better than 1/60 sec anyway) which i then split up and publish using sensor_msgs. I think sensor_msgs is the only way to do it as I need to set identical timestamps, but I'm a novice.

Also the publish rate is pretty slow. When I ask for 20 FPS I get about 5. message queue is set to 10. Would that be a problem? I suspect I'm going to have to redo it in C++ but I'd rather avoid that because python is quick to use.

Can someone tell me what I'm doing wrong, or a better way?

#!/usr/bin/env python

# picamera stereo ROS node using dual CSI Pi CS3 board
# Wes Freeman 2018
# modified from code by Adrian Rosebrock, pyimagesearch.com
# and jensenb, https://gist.github.com/jensenb/7303362

from picamera.array import PiRGBArray
from picamera import PiCamera
import time
import numpy as np
import cv2
import rospy
from sensor_msgs.msg import CameraInfo, Image
import yaml

# cam resolution
# x must be multiple of 128: 640, 512, 384
res_x = 384 # per camera
res_y = 288

# initialize the camera
print("Init camera...")
camera = PiCamera(stereo_mode = 'side-by-side',stereo_decimate=False)
camera.resolution = (res_x*2, res_y)
camera.framerate = 5
#camera.exposure_mode = 'sports'
camera.exposure_mode = 'antishake'
camera.video_stabilization = True
camera.awb_mode = 'tungsten'

rawCapture = PiRGBArray(camera, size=(res_x*2, res_y)) # picamera struct, a numpy array?

#setup the publishers
print("init publishers")
# queue_size should be roughly equal to FPS?
left_img_pub = rospy.Publisher('left_camera/image_color', Image, queue_size=10)
right_img_pub = rospy.Publisher('right_camera/image_color', Image, queue_size=10)
rospy.init_node('stereo_pub')

print("Setup done, entering main loop")
framecount=0
# capture frames from the camera
try:
    for frame in camera.capture_continuous(rawCapture, format="rgb", use_video_port=True):
        framecount +=1
        # grab the raw NumPy array representing the image and split it
        left_image = frame.array[:, :res_x, :]
        right_image = frame.array[:, res_x+1:, :]

        # clear the stream in preparation for the next frame
        rawCapture.truncate(0)

        stamp = rospy.Time.now()

        left_img_msg = Image()
        left_img_msg.height = res_y
        left_img_msg.width = res_x
        left_img_msg.step = res_x*3 # bytes per row: pixels * channels * bytes per channel (1 normally)
        left_img_msg.encoding = 'rgb8'
        left_img_msg.header.frame_id = 'image_raw'
        left_img_msg.header.stamp = stamp
        left_img_msg.data = left_image.flatten().tolist()

        right_img_msg = Image()
        right_img_msg.height = res_y
        right_img_msg.width = res_x
        right_img_msg.step = res_x*3
        right_img_msg.encoding = 'rgb8'
        right_img_msg.header.frame_id = 'image_raw'
        right_img_msg.header.stamp = stamp
        right_img_msg.data = right_image.flatten().tolist()

        #publish the image pair
        left_img_pub.publish(left_img_msg)
        right_img_pub.publish(right_img_msg)

        print("published frame: ", framecount)           



except rospy.ROSInterruptException:
    camera.close()
    pass

Revised code after advice in this thread:

#!/usr/bin/env python

# picamera stereo ROS node using dual CSI ...