cs4home_sound_project

Use case of the cs4home architecture prototype focused on the cognitive module of sound perception.

Installation

sudo apt install ros-humble-rclcpp-cascade-lifecycle*
sudo apt install ros-humble-ament-cmake-clang-format

cd ~/ros2_ws/src
git clone https://github.com/CoreSenseEU/cs4home_sound_project
vcs import --recursive < cs4home_sound_project/thirparty.repos
cd ~/ros2_ws
colcon build --symlink-install
ros2 launch cs4home_sound_project launch_sound.launch.py

Simulator

If needed, you can use the RB1 robot simulation:

git clone -b mic-array-urdf https://github.com/igonzf/ros2_rb1.git

Creating a Cognitive Module

Each cognitive module is defined through configuration and implemented using the base classes provided by the architecture.

1. Define the module in YAML

Edit the file:

cs4home_sound_project/config/params.yaml

Example:

sound_recognition:
  ros_parameters:
    core: sound_context_evaluation
    afferent: sound_input
    sound_input:
      topics: ["/audio", "/doa"]
      types:
        ["audio_common_msgs/msg/AudioData", "geometry_msgs/msg/PoseStamped"]
    efferent: sound_context_output
    sound_context_output:
      topics: ["/sound_context", "/context_markers"]
      types: ["sound_msgs/msg/SoundContext", "visualization_msgs/msg/Marker"]
    meta: sound_meta
    coupling: sound_coupling

Fields:

• core: name of the main processing node
• afferent: input handler (subscribes to topics)
• efferent: output handler (publishes results)
• meta: module metadata manager
• coupling: logic that connects with other modules

2. Implement the Module

Create a class that inherits from cs4home_core::CognitiveModule:

class SoundModuleCognitive : public cs4home_core::CognitiveModule {
// Instantiate lifecycle components here
};

Define each component as a subclass:

class AudioInput : public cs4home_core::Afferent {};
class AudioOutput : public cs4home_core::Efferent {};
class DefaultCoupling : public cs4home_core::Coupling {};
class AudioMeta : public cs4home_core::Meta {};

class SoundRecognition : public cs4home*core::Core {
void process() override {
    // Access afferent input by index
    auto msg_audio = afferent*->get_msg<audio_common_msgs::msg::AudioData>(0);

    // Or by topic name
    auto msg_doa = afferent_->get_msg<geometry_msgs::msg::PoseStamped>("/doa");

    // Publish processed result
    efferent_->publish(0, sound);

}
};

Each component is managed as a ROS 2 Lifecycle Node and instantiated according to the YAML configuration.