Skip to content

combine NeRF and Reinforcement Learning for Robot Navigation in indoor scenes

Notifications You must be signed in to change notification settings

danelpeng/NeRF-RL-Navigation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

environment

python环境

  1. conda create -n nerf-rl-nav python = 3.8
  2. pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt

仿真环境

测试环境

python test.py

复位环境

python reset.py ##启动gazebo和ros后 若非正常关闭,手动复位

训练VAE

激活工作空间

source devel/setup.bash

设置gazebo模型环境变量

export GAZEBO_MODEL_PATH=~/ros_project/catkin_ws/src/models

收集数据

  1. cd encoder
  2. python generate_data.py

训练

  1. python run.py

重建效果

加载保存的模型

vae_model = VAE(in_channels=3, latent_dim=128)	#初始化VAE
model_dict = torch.load("/checkpoint/**.ckpt", map_location=torch.device("cpu"))	#加载模型
new_state_dict = OrderedDict()	#消除模型key值不匹配
for k,v in state_dict.items():
	name = k[6:]
	new_state_dict[name] = v
	
vae_model.load_state_dict(new_state_dict)	#加载预训练模型,用于策略网络的前端

About

combine NeRF and Reinforcement Learning for Robot Navigation in indoor scenes

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published