# puppersim

**Repository Path**: codeinLinXu/puppersim

## Basic Information

- **Project Name**: puppersim
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-10-30
- **Last Updated**: 2021-10-30

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# puppersim
Simulation for DJI Pupper v2 robot

## Usage:

python setup.py develop

Then run puppersim/pupper_server.py

In a separate terminal, run the StanfordQuadruped run_djipupper_sim from this [fork](https://github.com/erwincoumans/StanfordQuadruped).

Keyboard controls:
* wasd: left joystick
* arrow keys: right joystick
* q: L1
* e: R1
* ijkl: d-pad
* x: X
* square: u
* triangle: t
* circle: c

## Training a Gym environment

You can train the pupper using pybullet [envs_v2](https://github.com/bulletphysics/bullet3/tree/master/examples/pybullet/gym/pybullet_envs/minitaur/envs_v2) and this [ARS fork](https://github.com/erwincoumans/ars).

```
pip install pybullet arspb ray puppersim
ray start --head
python puppersim/pupper_ars_train.py --policy_type=linear
python puppersim/pupper_ars_run_policy.py --expert_policy_file=data/lin_policy_plus_best_xxx.npz --json_file=data/params.json
```

See a video of a trained policy: https://www.youtube.com/watch?v=JzNsax4M8eg