The implementation of reinforcement learning Algorithm TD3 for robot arm in RLBench Environment.
Current implementation is on reach target task, more tasks will be added in the future.
 |
 |
|---|---|
| RLBench Environment | RLBench Gym Environment |
- Python 3.6
- Paddlepaddle >= 1.8.2
- PARL >= 1.3.1
- RLBench
- Pyrep
- CoppeliaSim (previously called V-REP)
RLBench is built around PyRep and CoppeliaSim. First head to the Pyrep github page and install them.
After that, go to RLBench github repo and install it .
- To start the training program with gym GUI
$ pip3 install -r requirements.txt
$ python train_gym.py --train --mode='human'
Render mode: human, rgb_array and None
To test agent
$ python train_gym.py --test --mode='human'
- To start the training program with RLBench GUI
$ python train_rlbench.py --train
$ python train_rlbench.py --test
You can also run it headless with argument --headless
- RLBench with Gym Env
- RLBench with Simple Task