jginesclavero/ros2_knowledge_graph

ros2_knowledge_graph

Many times, when you are developing a Software Architecture for robots, you need shared memory where to store information. There are many strategies for this. One of them is a blackboard. Another approach is a graph in which we store elements as graph nodes and relations as graph edges.

ROS2 knowledge graph provides you a way to share a graph between ros2 nodes running in an application. The graph is distributed in all the ros2 nodes. Each ros2 node contains a replica that is synchronized with all the other replicas, guaranteeing a Strong Eventual Consistency.

This video shows the functionality of this package:

To use it, you only have to create a instance of a ros2_knowledge_graph::GraphNode

#include "ros2_knowledge_graph/GraphNode.hpp"

class NodeA : public rclcpp::Node
{
public:
  NodeA()
  : rclcpp::Node("node_A")
  {
    graph_ = std::make_shared<ros2_knowledge_graph::GraphNode>("node_A");
    graph_->start();
  }

private:
  std::shared_ptr<ros2_knowledge_graph::GraphNode> graph_;
};


int main(int argc, char ** argv)
{
  rclcpp::init(argc, argv);
  auto node = std::make_shared<NodeA>();

}

The complete API of a ros2_knowledge_graph::GraphNode is at ros2_knowledge_graph/include/ros2_knowledge_graph/GraphNode.hpp, but it is mainly:

  bool add_node(const Node & node);
  bool remove_node(const std::string node);
  bool exist_node(const std::string node);
  boost::optional<Node> get_node(const std::string node);

  bool add_edge(const Edge & edge);

  bool remove_edge(const Edge & edge);
  bool exist_edge(const Edge & edge);
  void get_edges(
    const std::string & source, const std::string & target, const std::string & type,
    std::vector<Edge> & result);

I use Pluginlib to add special funtionality to nodes and edges depending on its type. Check Plugins for an example.

Have fun!!!