# LearningHumanoidRunning

**Repository Path**: chengfeng2015_bolt/LearningHumanoidRunning

## Basic Information

- **Project Name**: LearningHumanoidRunning
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: BSD-2-Clause
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-06-07
- **Last Updated**: 2026-06-07

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# LearningHumanoidRunning

![climb_down](https://github.com/user-attachments/assets/0cbca7ab-ade9-4e77-9b5c-9d2fafead47f)



Training humanoid robots to run using reinforcement learning, modified from the work at https://github.com/rohanpsingh/LearningHumanoidWalking, with additions including: 
1. Running using only legs.
2. Using hands for balance.
3. Running with both hands and legs simultaneously.

Added 14 new arm joints, increasing the observation dimension from 37 to 65, and added some reward functions related to arm movements to support the training of robotic arm operations.

## Code structure:
A rough outline for the repository that might be useful for adding your own robot:
```
LearningHumanoidWalking/
├── envs/                <-- Actions and observation space, PD gains, simulation step, control decimation, init, ...
├── tasks/               <-- Reward function, termination conditions, and more...
├── rl/                  <-- Code for PPO, actor/critic networks, observation normalization process...
├── models/              <-- MuJoCo model files: XMLs/meshes/textures
├── trained/             <-- Contains pretrained model for JVRC
└── scripts/             <-- Utility scripts, etc.
```

## Requirements:
- Python version: 3.7.11  
- [Pytorch](https://pytorch.org/)
- pip install:
  - mujoco==2.2.0
  - [mujoco-python-viewer](https://github.com/rohanpsingh/mujoco-python-viewer)
  - ray==1.9.2
  - transforms3d
  - matplotlib
  - scipy

## Usage:

Environment names supported:  

| Task Description                | Environment name |
|---------------------------------|------------------|
| Basic Walking Task              | 'jvrc_walk'      |
| Stepping Task (using footsteps) | 'jvrc_step'      |
| Walking Task (using arm)        | 'jvrc_arm'       |
| run Task (only using leg)       | 'jvrc_run'       |
| run Task (using leg and arm)    | 'jvrc_run_arm'       |


#### **To train:** 

```
$ python run_experiment.py train --logdir <path_to_exp_dir> --num_procs <num_of_cpu_procs> --env <name_of_environment>
```  


#### **To play:**

We need to write a script specific to each environment.    
For example, `debug_stepper.py` can be used with the `jvrc_step` environment.  
```
$ PYTHONPATH=.:$PYTHONPATH python scripts/debug_stepper.py --path <path_to_exp_dir>
```

#### **What you should see:**





https://github.com/user-attachments/assets/08628f41-29f4-463e-947a-f9cd4d0b210c