rl-obstacle-avoidance
rl-obstacle-avoidance copied to clipboard
LIDAR based Obstacle Avoidance with Reinforcement Learning
LIDAR based Obstacle Avoidance with Reinforcement Learning
This project lists all of the deliverables for the TUM project course Applied Reinforcement Learning (Summer Semester 2019).
Results
Reward Graphs:
State Representation | Linear Value Function Approximation | Algorithms |
---|---|---|
Project Demo.:
Simulation | Real Turtlebot |
---|---|
![]() |
![]() |
Supplementary Material:
Requirements
- Python 2.7 (Python 3 for sensor-model fitting and auto-encoder training)
- ROS-Kinetic with turtlebot
- Catkin
- PyTroch
- Scipy
- PyYAML
Instructions
- Move the
rl_tb_lidar
andstage_ros_u
folders tocatkin_ws/src
directory. - run
catkin_make
in thecatkin_ws
directory. - Run
source devel/setup.bash
command in thecatkin_ws
directory. - Run
roslaunch rl_tb_lidar tb_stage_m1.launch
to launch only stage. - Open an another terminal, go to the directory of the python script e.g.
cd ~/catkin_ws/src/rl_tb_lidar/src
and runpython main.py configs/config.yaml
. - To try different configurations, edit the
configs/config.yaml
file accordingly.
Versioning
We version the project with each new deliverable. For the versions available, see the tags on this repository.
Authors
- Akbar, Uzair - [email protected]
- Gundogan, Alperen - [email protected]
- Ellouze, Rachid - [email protected]
See also the list of contributors.