thecamusean/sh8_prediction repository overview

Initialize the docker container. In a Terminal,

''sudo docker run -it --name test thecamusean/sh8_prediction bash''

Inside the Container, initialize ROS

''roscore''

---

in a different terminal, access to the container:

''sudo docker exec -it test bash''

and load the rosbag file for testing:

'' cd src/rosbags/ && rosbag play -l sh8_file.bag''

---

In a new Terminal, enter in the container

''sudo docker exec -it test bash''

And start the ROS server:

'' cd src/eval_logp/scripts && python3 eval_log_prob_cont_simple_train_server_gmm.py''

This ROS node will initialize a ros service in which we can request the human likelihood in a certain region

---

Finally, we evaluate the likelihood by a ROS service client. Enter in the container:

''sudo docker exec -it test bash''

And test the client:

'' cd src/eval_logp/scripts && python3 eval_log_prob_client.py ''

---

START/STOP training

We have added an additional topic to control the continuous training. You can initialize the training by

To init the continuous training:

rostopic pub /motion_prediction_train_control std_msgs/String "data: 'START'"

To stop the continuous training:

rostopic pub /motion_prediction_train_control std_msgs/String "data: 'STOP'"

---

You can visualize the occupancy of the conditioned human occupancy by

python3 eval_log_prob_client_env.py

It generated a topic publisher /visualization_marker_array of MarkerArray that can be visualized in Rviz.