I have a problem with running the simulator with only image data. It complains about
E0716 19:03:37.176589 29053 lane_post_processing_subnode.cc:154] failed to get shared data. event:event_id: 0 timestamp: 0.000000000 reserve:
E0716 19:03:37.176601 29053 lane_post_processing_subnode.cc:196] Failed to get shared data. event:event_id: 0 timestamp: 0.000000000 reserve:
It seems camera_object_data_->Get(data_key, objs) is NULL.
According to readme at https://github.com/ApolloAuto/apollo/tree/master/modules/perception
perception modules should publish at rostopic echo /apollo/perception/obstacles
When running ./scripts/perception.sh everyting is ok. rostopic echo /apollo/perception/obstacles receives data. When running ./scripts/perception_lowcost.sh nothins is shown in dreamview, or rostopic echo /apollo/perception/obstacles.
Any ideas? We want camera only detection, no lidar involved.
Upade:
One possible problem is that when I play this bag:
rosbag play demo-sensor-data-apollo-2.0/2018-01-03-19-37-16.bag
It doesn't publish data to
rostopic echo /apollo/sensor/camera/obstacle/front_6mm
In fact, rostopics the bag has are:
rosbag info demo-sensor-data-apollo-2.0/2018-01-03-19-37-16.bag
topics: /apollo/canbus/chassis 5851 msgs : pb_msgs/Chassis
/apollo/localization/pose 5856 msgs : pb_msgs/LocalizationEstimate
/apollo/sensor/camera/traffic/image_long 471 msgs : sensor_msgs/Image
/apollo/sensor/camera/traffic/image_short 469 msgs : sensor_msgs/Image
/apollo/sensor/conti_radar 789 msgs : pb_msgs/ContiRadar
/apollo/sensor/gnss/best_pose 59 msgs : pb_msgs/GnssBestPose
/apollo/sensor/gnss/corrected_imu 5838 msgs : pb_msgs/Imu
/apollo/sensor/gnss/gnss_status 59 msgs : pb_msgs/GnssStatus
/apollo/sensor/gnss/imu 11630 msgs : pb_msgs/Imu
/apollo/sensor/gnss/ins_stat 118 msgs : pb_msgs/InsStat
/apollo/sensor/gnss/odometry 5848 msgs : pb_msgs/Gps
/apollo/sensor/gnss/rtk_eph 49 msgs : pb_msgs/GnssEphemeris
/apollo/sensor/gnss/rtk_obs 352 msgs : pb_msgs/EpochObservation
/apollo/sensor/velodyne64/compensator/PointCloud2 587 msgs : sensor_msgs/PointCloud2
/tf 11740 msgs : tf2_msgs/TFMessage
/tf_static 1 msg : tf2_msgs/TFMessage
From pereception doc:
The perception module inputs are:
Radar data (ROS topic /apollo/sensor/conti_radar)
Image data (ROS topic /apollo/sensor/camera/obstacle/front_6mm)
I tried to modify: /apollo/modules/perception/data/params/start_obstacle_camera.launch
<launch>
<arg name="front_obstacle_6mm_topic_name" default="/apollo/sensor/camera/obstacle/front_6mm" />
<arg name="frame_rate" default="30"/>
to
<launch>
<arg name="front_obstacle_6mm_topic_name" default="/apollo/sensor/camera/traffic/image_long" />
<arg name="frame_rate" default="30"/>
But it doesn't work.
Many thanks!
Vlad
You should be able to download demo_2.5.bag
Thank you so much!
Steps followed:
rosbag info demo_2.5.bag
/apollo/perception/obstacles - needs to be removed/apollo/sensor/camera/obstacle/front_6mmrosbag filter demo_2.5.bag demo_2.5_cam.bag "topic != '/apollo/perception/obstacles'"
rosbag info demo_2.5_cam.bag doesn’t contain detected objects anymore../scripts/perception_lowcost.sh start
rosbag play demo_2.5_cam.bag --loop
Got image data, and detection in:
rostopic echo /apollo/perception/obstacles
rostopic echo /apollo/sensor/camera/front_6mm
Unfortunately Dreamview output looks empty - no image, no map, no lane lines.
However, PerceptionObstacle module is responding (small delay).
Unfortunately rostopic echo /apollo/perception/obstacles shows no lane marker when running ./scripts/perception_lowcost.sh :
header {
timestamp_sec: 1531779145.71
module_name: "perception_obstacle"
sequence_num: 77
lidar_timestamp: 0
camera_timestamp: 1523917555000912128
radar_timestamp: 0
}
error_code: PERCEPTION_ERROR_PROCESS
lane_marker {
}
LanePostProcessing node lane_post_processing_subnode.cc still fails at line 154
AERROR << "failed to get shared data. event:" << event.to_string();

```
Does the visualizer work?
./scripts/perception_offline_visualizer.sh
It took me a long time before I was able to get this to work, but I'll see if I am any help.
No, I couldn't run ./scripts/perception_offline_visualizer.sh becuase of docker OpenGL error.
Unfortunately rostopic echo /apollo/perception/obstacles shows no lane marker when running ./scripts/perception_lowcost.sh :
header {
timestamp_sec: 1531779145.71
module_name: "perception_obstacle"
sequence_num: 77
lidar_timestamp: 0
camera_timestamp: 1523917555000912128
radar_timestamp: 0
}
error_code: PERCEPTION_ERROR_PROCESS
lane_marker {
}
yea, the openGL error is a headache. I think Nvidia drivers need to be installed with a flag: --no-opengl-files.
But ignoring the visualizer,
To verify the perception module is working... I would create a new bag file that contains images but not detections... so only this topic, /apollo/sensor/camera/front_6mm. Then play that new bag, and run the perception module without visualization. see if you are able to rostopic echo /apollo/perception/obstacles
Thank you! That's precisely what I did.
Here is the update
/apollo/modules/perception/conf/perception_lowcost.conf--dag_config_path=conf/dag_camera_obstacle.config # only camera and objectsNo lane postprocessing node.
It outputs at rostopic echo /apollo/perception/obstacles
perception_obstacle {
id: 8
position {
x: 45.1654319763
y: -5.86641836166
z: 0.0
}
theta: -0.0499055683613
Also, I'm able to see something when selecting Navigation/Mkz Example/Demo in Dreamview.
How can I do lane segmentation and OD? LanePostProcessing Node fails when I run it.
Do you know how is it doing 3D bouding boxes using yolo?

"How can I do lane segmentation and OD?"
I am not sure what this means.
LanePostProcessing Node fails when I run it.
This works for me: https://github.com/ApolloAuto/apollo/blob/master/modules/perception/conf/dag_camera_obstacle_lane_vis.config
and you can even modify it to remove the visualizer, so its just camera and lanes.
"Do you know how is it doing 3D bouding boxes using yolo?"
I am not sure, I'd like to know.
I wish there was a better sample bag file with images. I can upload and share one later this week.
Demo2.5 is not a very good sample IMO.
Most helpful comment
You should be able to download demo_2.5.bag