NaovaCodeK1 Installation

NaovaCodeK1 Installation#

This page explains how to install, build, and run NaovaCodeK1.

Note

Target platform: Ubuntu 22.04 (Jammy).

Prerequisites#

You need:

  • A Linux machine with Ubuntu 22.04

  • A user account with sudo permissions

  • Internet access (for package downloads)

  • Around 20+ GB free disk space (more if installing vision dependencies)

  • ROS2 Humble already installed

Install core build tools first:

sudo apt-get update
sudo apt-get install -y python3-colcon-common-extensions python3-rosdep
sudo rosdep init
rosdep update

Install required system libraries:

sudo apt-get install -y \
        libopencv-dev \
        libpcl-dev \
        libeigen3-dev \
        libyaml-cpp-dev \
        libpcap-dev \
        ros-humble-backward-ros

1. Install Booster Robotics SDK#

Follow the official Booster Robotics SDK installation guide :

cd ~/naova/
git clone https://github.com/BoosterRobotics/booster_robotics_sdk.git
cd booster_robotics_sdk
sudo ./install.sh

2. Clone NaovaCodeK1 repository#

If you do not have the repository yet:

cd ~/naova/
git clone git@github.com:Naova/NaovaCodeK1.git
cd NaovaCodeK1

3. Install extra dependencies#

Install remaining ROS package dependencies from this workspace:

source /opt/ros/humble/setup.bash
rosdep install --from-paths src --ignore-src -r -y

For build without CUDA (ONNX inference), install ONNX Runtime first.

On aarch64:

cd ~/naova/NaovaCodeK1
# aarch64
./third_party_aarch64/install_onnxruntime.sh

On x86_64:

cd ~/naova/NaovaCodeK1
# x86_64
./third_party/install_onnxruntime.sh

Build#

Important

Most users should use the without CUDA flow unless they have a GPU (Nvidia GeForce RTX 5050 or higher) that supports CUDA Compute Capability 12.0 or higher.

Build without CUDA (requires ONNX Runtime):

./scripts/build_no_cuda.sh

Build with CUDA (real robot):

./scripts/build.sh

Shell environment (important):

In each new terminal, source ROS first, then this workspace overlay:

source /opt/ros/humble/setup.bash
source install/setup.bash

Notes:

  • If you only source /opt/ros/humble/setup.bash, this workspace packages are not overlaid.

  • After any colcon build, run source install/setup.bash again in that terminal.

Run#

Simulation (virtual robot):

For first-time setup, copy the example local override files once:

cp src/brain/config/config_local.example.yaml src/brain/config/config_local.yaml
cp src/vision/config/vision_local.example.yaml src/vision/config/vision_local.yaml

Then start simulation:

./scripts/sim_start.sh

Local runs use the committed local override files directly:

  • src/brain/config/config_local.yaml

  • src/vision/config/vision_local.yaml

The launch files load the base config first, then the local override file, and finally the machine-specific files in ~/agents/booster_soccer/ if they exist. This keeps the team-shared local defaults in git while still allowing robot-specific values on the hardware.

Real robot:

./scripts/start.sh

Real robot runs use the same launch flow, but the local override files contain the real-robot values instead of the simulation values.

Configuration notes#

Config precedence:

  1. config.yaml / vision.yaml

  2. config_local.yaml / vision_local.yaml

  3. ~/agents/booster_soccer/brain.yaml / ~/agents/booster_soccer/vision.yaml

JetPack 6.2 note:

This repository supports JetPack 6.2 and is adapted to the default TensorRT model configured in src/vision/config/vision.yaml.

detection_model:
        model_path: ./src/vision/model/best_digua_second_10.3.engine
        confidence_threshold: 0.2