Regarding Map Generation

[AdithyaIniesta]

Dear Sir/Madam,

I executed all the commands that is listed in read me section. Training too was successful. Currently I do not know how to generate the map. Any documentation on this will help me a lot.

Thank you very much for timely help.

Kind regards,
Adithya

[soraKING44]

Thank you for contacting me.
I am the first author of this research and a maintainer for this repository.

You mean that you want to generate semantic maps?
"colormap_saver.cpp" is the code for generating semantic maps, and it is called from "spco_mapping.sh".

In the "colormap_saver.cpp", could you check the "std::string mapname" on L119.

spco_mapping/src/colormap_saver.cpp

Line 119 in b4a31a7

std::string mapname = "PATH/catkin_ws/src/spco_mapping/data/" + hoge;

I cannot confirm about that because Ubuntu14.04 is too old for me. But, maybe it can help you. I'm sorry, I forgot to explain about save file path on README.

[AdithyaIniesta]

Dear Sir,

Thank you very much for speedy response. I'm currently working on my thesis and I have shortlisted the proposed method for benchmark analysis. It would be of great help if you could provide me with complete list of commands in README section.

Yours timely help will boost my research work.

Kind regards,
Adithya

[soraKING44]

Before running "./spco_mapping.sh", please change "colormap_saver.cpp" L119 to your path (if you generated catkin_ws with roswiki, change "PATH" to your home dir path). You need catkin_make.
Semantic maps are saved in "$(your setting path)/catkin_ws/src/spco_mapping/data/$(trialname)/*.ppm".

I think it will work. If you have any trouble, please tell me your error.

[AdithyaIniesta]

Dear Sir,

I did all the corrections in path you have mentioned. '

When I run "roslaunch spco_mapping amcl.launch" I get this warning "[ WARN] [1602773227.022408028]: No laser scan received (and thus no pose updates have been published) for 1602773227.022250 seconds. Verify that data is being published on the /scan topic"

and after run "./spco_mapping.sh" I did rostopic list. For which I got
/amcl/parameter_descriptions
/amcl/parameter_updates
/amcl_pose
/initialpose
/map
/map_metadata
/particlecloud
/rosout
/rosout_agg
/scan
/spco/color_map
/spco/start_learning
/tf
/tf_static

There is no topic called "/spco/read_dataset" to run the command "rostopic pub /spco/read_dataset std_msgs/Int8 1".

In $trialname folder I got folders such as pi theta w along with init.py. There was no .ppm file.

Also rqt_graph does not show up. When I run other packages, I can see rqt_graph.

Please help me resolve this issue.

Kind regards,
Adithya

[soraKING44]

OK, could you check the output message?

If some directories are generated, "mkdir + DATASET_FOLDER + TRIALNAME" can be confirmed on terminal, right? (in the do_mkdir() function of learning.py L28)
do_mkdir() function is called from execute() function. It is callback function of "/spco/start_learning".
In the execute() function, read_dataset() function and learning() function are also called. You can confirm "start read_dataset" and "start learning" as message on terminal.
If the learning() function is called correctly, there is no problem. Learning should start and semantic maps will be generated in save_parameter() function on L195.
Is there some error message?

I am worried about get_map_callback() function is not called. Could you try to change L218,
rospy.Subscriber("/spco/start_learning", String, self.get_map_callback, queue_size=1)
↓
rospy.Subscriber("/spco/read_dataset", String, self.get_map_callback, queue_size=1)

[AdithyaIniesta]

Dear Sir,

I got an error when I changed L218 as per your instruction. Prior to this change in the terminal I was able to see the message
start read_dataset
start learning.
saved color_map.

I'm still not able to generate and save the map. Sir please help me to figure out the issue.

[soraKING44]

Please check the PATH in the save_parameter() function.
And, could you show me what error appears?

if you can, please confirm "rostopic echo /spco/color_map" is sended on another terminal.

[AdithyaIniesta]

Dear Sir,

In colour.launch, there was a spelling errror of Trailname which I corrected. I was able to see * /spco/read_dataset [std_msgs/Int8] 1 publisher amoung published topic.

I understood the flow of learning.py and I was able to see "start read_dataset" and "start learning" on the terminal as well. Then it took around one hour for training and finally displayed the message "finish learning".

Also when I run rviz, I was able to one of the occupancy grid maps located in the dataset folder. RQT Graph is also

So far good.

visible.

But I didn't understand how colormap_saver.cpp is called.. This file generates .ppm file as per my understanding. The cpp file is not executed by colour_saver launch file and I think this is where I'm not able to get the output.

I think I'm close to getting output but I just need little more guidance from you.

Hoping for quick response. @soraKING44

Kind regards,
Adithya

[soraKING44]

colormap.launch calls colormap_saver.cpp.
If color map saver is called, message "Waiting for the map" is shown on terminal (L51). Can you confirm?
If no, please confirm package name and catkin_make.

[AdithyaIniesta]

I manually ran roslaunch colorsaver and indeed got the messgae waiting for the map. When learning starts, I saw message .ppm files generated in the folder. This is the sample image from an iteration.

This map does not look like the one shown in the paper. Not sure where I'm going wrong.

[soraKING44]

That looks like correct result.
Just not enough iteration. Please iterate 5000~10000 times for learning.

[AdithyaIniesta]

Sir, I have a doubt, the commands in .sh file
gnome-terminal --command 'python learning.py '$trialname and roslaunch spco_mapping colormap.launch hoge:=$trialname
should they run at the same time or one after the other?. And to my disspointment, I got 5000 .ppm images in the directory with all images looking like

Any mistake I doing Sir?
I opened five terminals in terminator.
Term1 : roslaunch spco_mapping amcl.launch
Term2: roslaunch spco_mapping colormap.launch hoge:=$trialname
Term3: ./spco_mapping.sh
Term4: rostopic pub /spco/read_dataset std_msgs/Int8 1
Term5: rostopic pub /spco/start_learning std_msgs/String "start"

I just place one room "room1dk5" folder in "catkin_ws/src/spco_mapping/data/DATASET".

The results are not satisfactory as shown in the paper.

[soraKING44]

Are the parameters correct?
If MRF is weak, please set more large gamma. Also, the other parameter might need to adjust.