About installation problems
Opened this issue · 11 comments
Excuse me, I read your papers and it is a great work, but I have some doubts about it.
1.Does it support Ubuntu version>16.04?
2.Are all the installation steps related in remote PC? Is there any installation step in turtlebots? It is not so clear.
Excuse me, I read your papers and it is a great work, but I have some doubts about it.
1.Does it support Ubuntu version>16.04?
2.Are all the installation steps related in remote PC? Is there any installation step in turtlebots? It is not so clear.
Hello, I haven't tried ubuntu 18 yet, but I think it should work, you can refer to this link: https://emanual.robotis.com/docs/en/platform/turtlebot3/quick-start/. You can choose ubuntu 18.04 and choose melodic there. You can also find the installation step for turtlebot there, hope this reply works for you!
Thanks! Well, I still have some doubts.
1.Are all computationally intensive works on the PC? And turtlebots just collect data, receive cmd from PC?
2.Or is there any workload balancing between turtlebots and PC?
3.I saw the video and find that we should train it in the gazebo with a human in the loop? Does it mean we should sit in a chair and play for hours?
video and find that we should train it in the gazebo with a human in the loo
For your questions, 1 & 2, all the intensive works are on the PC. 3, human data collection is done in around 20 minutes
Thanks a lot! I will have a try,.
BYW, is there any model that has been trained provided?
Thanks a lot! I will have a try,.
BYW, is there any model that has been trained provided?
Yest, it is in the script files
Thanks! I try to install it but comes with some problems
-
Just play with the trained model(provided, human including) in real world
-
Params
PC: 192.168.1.2(master) Single Turtlebot3: 192.168.1.3
-
Step1
Set the network
-
Step2
PC:Command1: roscore Into virtual env(it is tf2 using conda here, it can work) Command2: (dir: turtlebot3_ddpg_collision_avoidance/turtlebot_ddpg/launch) roslaunch turtlebot3_laser_filter.launch(filter laser data) Command3: (dir: turtlebot3_ddpg_collision_avoidance/turtlebot_ddpg/scripts/fd_replay/play_human_data/) set train_indicator=0 under ddpg_network_turtlebot3_amcl_fd_replay_human.py python ddpg_network_turtlebot3_amcl_fd_replay_human.py(sub laser data and return result with the trained model)
Turtlebot3:
Command4:
roslaunch turtlebot3_bringup turtlebot3_robot.launch multi_robot_name:="tb3_0" set_lidar_frame_id:="tb3_0/base_scan"
But it reports error when running command2 & 3
command2--error1
command3--error2
I don't know what happens for commands 2&3, is it the step problem? Or I just ignore sth else?
Looking forward to receiving your reply
I have solved the second error, too. It seems that I should run the model at the end, then it works. But the turtlebot3 doesn't move, here are the details:
The commands are the same as we mentioned before.
When running the Command3 at the end: (dir: turtlebot3_ddpg_collision_avoidance/turtlebot_ddpg/scripts/fd_replay/play_human_data/)
set train_indicator=0 under ddpg_network_turtlebot3_amcl_fd_replay_human.py
python ddpg_network_turtlebot3_amcl_fd_replay_human.py(sub laser data and return result with the trained model)
It reports like this:
And it rqt_graph, I find that the filter and mode nodes are not active:
I think maybe I ignore some steps.
- Can you modify your detailed commands in the readme.md? For example, how to start the turtlebot3 board? I don't whether I am right. How to run the filter node? ...
- Can you add some details about some tools' version you use? For example, TensorFlow, Keras, python... I think this is very important.