This project enables web-based teleoperation support for different types of robots like MiniPupper. Video streaming is implemented via WebRTC, and teleoperation is done via ROS. Note that the backend expects you've already connected a depth camera like OAK-D Lite to your robot (only Luxonis products are supported at the moment). If you don't have one yet, you can still teleoperate the robot via keyboard w/o a video stream, or just use a simulator (guide TBD).

• RoboPerception
  • Architecture
  • Installation
  • Running
  • Building FE and BE Docker images
  • ToDo

The following diagram reflects the most recent implementation:

• backend: Python BE, which streams camera video to the remote browser via WebRTC; it internally uses DepthAI API on real hardware, and Gazebo camera images in simulation mode;
• frontend: ReactJS FE which uses roslibjs to communicate with ROS bridge and WebRTC API for camera streaming;
• rosbridge: WS proxy between FE/BE and ROS that handles the following messages.

• ROS Bridge should be visible to a client to perform common operations via Web UI.
• You should provide CompressedImage to see a virtual camera stream from Gazebo in simulation mode.
• Only BatteryState.percentage property is used for rendering for battery stats.
• Teleoperation is impossible while is_standing or teleop_status flag is false.
• Keyboard keys are sent in a raw format. You should implement ROS subscriber which transforms a key to cmd_vel.
• /robot_pose/change is very robot-specific: FE just sends either a sit or stand command.

Clone the source code:

git clone https://github.com/WaverleySoftware/robo-perception.git && cd robo-perception

Prepare required env files:

./generate_configs.sh [ROBOT_IP_ADDRESS]

Robot's IP is required for the FE to be able to communicate with the BE via remote browser.

Setup backend:

cd backend && python3 -m venv .venv
source .venv/bin/activate
pip3 install pip --upgrade
pip3 install -r requirements.txt

Setup frontend:

cd ../frontend
npm install

Start ROS bridge and all the custom ROS services that implement the message protocol described in architecture section.

Start backend:

cd robo-perception/backend && source .venv/bin/activate
./run.sh

Start frontend:

cd robo-perception/frontend && npm start

Open your web browser and go to: http://[ROBO_PERCEPTION_SERVICE_IP_ADDRESS]:3000

Run the following command to build FE and BE images:

docker compose build

• Polish FE and BE code
• Add local deployment instructions
• Add simulated environment instructions
• Add Docker instructions
• Migrate to ROS2
• Add interactive map for SLAM and navigation
• Add gamepad support