Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/sskorol/minipupper-teleop
MiniPupper robot teleoperation via ROS and WebRTC
https://github.com/sskorol/minipupper-teleop
champ depthai docker docker-compose fastapi minipupper mobx mui noetic oak-d pydantic reactjs robotics ros rosbridge roslibjs streaming teleop teleoperation webrtc
Last synced: 26 days ago
JSON representation
MiniPupper robot teleoperation via ROS and WebRTC
- Host: GitHub
- URL: https://github.com/sskorol/minipupper-teleop
- Owner: sskorol
- License: mit
- Created: 2022-08-07T23:23:55.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2023-09-15T22:55:00.000Z (about 1 year ago)
- Last Synced: 2023-09-16T00:17:43.922Z (about 1 year ago)
- Topics: champ, depthai, docker, docker-compose, fastapi, minipupper, mobx, mui, noetic, oak-d, pydantic, reactjs, robotics, ros, rosbridge, roslibjs, streaming, teleop, teleoperation, webrtc
- Language: JavaScript
- Homepage:
- Size: 10.8 MB
- Stars: 7
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
## MiniPupper Teleop
![Build Status](https://github.com/sskorol/minipupper-teleop/actions/workflows/main.yml/badge.svg?branch=main)
This project allows streaming video from [MiniPupper](https://minipupperdocs.readthedocs.io/en/latest/) via WebRTC and teleoperating it via ROS. Note that the [backend](https://github.com/sskorol/minipupper-teleop/tree/main/backend) expects you've already connected [OAK-D Lite](https://shop.luxonis.com/products/oak-d-lite-1) camera to your robot. If you don't have one yet, you can still teleoperate the robot via keyboard, but w/o a camera stream. Also, note that technically you are not forced to use this project with MiniPupper only. It should work for any robot with OAK camera.
- [MiniPupper Teleop](#minipupper-teleop)
- [Architecture](#architecture)
- [Installation](#installation)
- [Running on MiniPupper](#running-on-minipupper)
- [Simulated Environment](#running-in-simulated-environment)
- [Building FE and BE](#building-fe-and-be)
- [Known Issues](#known-issues)
- [ToDo](#todo)### Architecture
The following diagram reflects the most recent implementation:
![RoboticsDiagram](https://user-images.githubusercontent.com/6638780/184701436-863ccb48-9bcd-44c4-ba8f-c6ebf35d10d9.png)
- **roscore**: a master node that handles all the ROS requests;
- **rosbridge**: WS proxy between FE and ROS that accepts messages (keys) from the remote browser and passes them to ROS nodes;
- **webrtc-be**: Python BE, which streams OAK-D Lite camera video to the remote browser via WebRTC (internally uses [DepthAI](https://docs.luxonis.com/projects/api/en/latest/index.html) API);
- **teleop-fe**: ReactJS FE which uses [roslibjs](https://github.com/RobotWebTools/roslibjs) to communicate with ROS bridge and WebRTC API for camera streaming;
- **servo-drv**: MiniPupper's servo driver node, which listens to CHAMP messages and changes joints' angles;
- **CHAMP/vel-smoother**: [CHAMP](https://github.com/chvmp/champ) framework controls the robot's movements;
- **teleop**: a slightly modified [teleop-legged-robots](https://github.com/SoftServeSAG/teleop_legged_robots) node that accepts remote keys rather than local keyboard events.Note that `velocity-smoother` was intentionally splitted from `champ` due to netwroking issue mentioned in a [known issues](#known-issues) section.
### Installation
The following [ROS image](https://drive.google.com/file/d/1Mk_bSmIvnN8EIzB8IilS9M4pofTUH9r2/view?usp=sharing) already comes with all the required drivers pre-installed. Just flash it to SD card and you are almost ready to go.
Check the [official guide](https://docs.docker.com/engine/install/ubuntu/) if you don't have Docker yet. Note that you need both `docker` and `docker-compose` CLI tools installed on MiniPupper.
Clone the source code:
```shell
git clone https://github.com/sskorol/minipupper-teleop.git && cd minipupper-teleop
```Prepare calibration and env files:
```shell
./generate_configs.sh [MINI_PUPPER_IP_ADDRESS]
```Adjust angles in `calibration_settings.yaml` to match your own robot's calibration data. Note that MiniPupper's legs should be calibrated to 90 degress as on the following screenshot, as CHAMP framework automatically adjusts angles during bringup process to make your robot stand:
![image](https://user-images.githubusercontent.com/6638780/183618832-c133ddef-484c-4974-b6e9-04f7e1d81e6e.png)
MiniPupper's IP is required for the FE container to be able to communicate with the BE via remote browser.
Use the following steps to relax your web-browser restrictions:
- Open Chrome
- Type `chrome://flags/` in the address bar and hit Enter
- Enable `Insecure origins treated as secure` option, and type the IP address of your MiniPupper
- Restart Chrome### Running on MiniPupper
Run the following command to start a stack of docker images required to perform teleoperation:
```shell
docker compose pull && docker compose up -d
```An old docker cli uses a bit different syntax: `docker-compose up -d`.
Open your web browser and go to: `http://[MINI_PUPPER_IP_ADDRESS]`
### Running in simulated environment
If you don't have a robot yet, you can still play with teleoperation locally in a simulated environment.
```shell
# Required for running Gazebo in container
xhost +local:docker
# Download images
docker compose -f docker-compose-sim.yaml pull
# Run services required for simulation
docker compose -f docker-compose-sim.yaml up -d
```Then open your web-browser on a localhost, wait until teleop is ready, and you're good to go.
### Building FE and BE
Run the following command on MiniPupper to build FE and BE images:
```shell
docker compose build
```### Local deployment
Install backend dependencies:
```shell
cd ./backend && python3 venv .venv
source .venv/bin/activate && pip3 install pip --upgrade && pip3 install -r requirements.txt
```Start backend:
```shell
./run.sh
```Install frontend dependencies:
```shell
cd ../frontend && npm install
```Prepare `.env` with required environment variables:
```shell
cp .env.example .env
```Here are sample values:
```properties
REACT_APP_ROSBRIDGE_SERVER_IP=localhost
REACT_APP_ROSBRIDGE_SERVER_PORT=9090
REACT_APP_RECONNECTION_TIMER=1000
REACT_APP_BE_URL=http://localhost:8080
REACT_APP_IS_SIMULATION=true
```Start frontend:
```shell
npm run start
```Go to [http://localhost:3000](http://localhost:3000) to see the web UI.
Note that frontend and backend depend on the ROS bridge and a custom teleop node. Make sure you've started them beforehand.
### Known issues
In rare cases `teleop`, `smoother` and `servo` nodes can't correctly publish/subscribe to `/cmd_vel` topic due to registration failure. The current workaround is displayed on the following diagram.
You can try to restart docker images to see if it helps. I couldn't yet found the root cause of these Docker <---> ROS networking issues. Feel free to contact [author](mailto:[email protected]) if you have any idea on how to stabilize it.
Run the following command on MiniPupper to diagnose potential errors in logs:
```shell
docker compose logs -f
```### ToDo
- [x] Polish FE code
- [ ] Polish BE code
- [x] Add local deployment instructions
- [ ] Add docker-cross builds
- [ ] Push teleop and mini-pupper core sources
- [ ] Migrate to ROS2
- [ ] Get rid of velocity-smoother, which seems to cause most networking issues
- [ ] Add map for SLAM and navigation
- [ ] Potentially integrate this code into MiniPupper repo