File server.cpp#

Definition (server/src/server.cpp)#

Includes#

  • ExecutionManager.h

Namespaces#

Functions#

Full File Listing#

namespace RPC#

API functions for external communication with the MAPF planner and the executor (the robots).

namespace rpc_api#

Functions

string getRobotsLocation()#

Get the latest MAPF instance, newly finished task IDs, and initialization status.

Returns:

A JSON string with the following schema. 

{
  "initialized": true,
  "mapf_instance": {
    "starts": [ State, ... ],
    "goals":  [ [ Task, ... ], ... ]
  },
  "new_finished_tasks": [ 0, 1, 2, 3 ]
}

  • initialized indicates whether the system has completed initialization.

  • mapf_instance contains the current MAPF problem state:

    • starts is a list of State objects describing the current start states of all agents:

      • Each State has the structure: 

        {
  "location": 42,
  "timestep": 10,
  "orientation": 1
}

      • location is the flattened cell ID.

      • timestep is the start timestep, usually 0.

      • orientation is encoded as:

        • 0: East

        • 1: North

        • 2: West

        • 3: South

    • goals is a list of task lists, where each task list corresponds to a robot:

      • Each Task object has the structure: 

        {
  "id": 0,
  "location": 84,
  "task_wait_time": 0,
  "orientation": -1
}

      • id is the unique task identifier.

      • location is the flattened cell ID of the task location.

      • task_wait_time is the number of timesteps the agent should wait at the task location, usually 0.

      • orientation is the required orientation at the task location, with -1 indicating no constraint.

  • new_finished_tasks is a list of task IDs completed since the last query.

void addNewPlan(string &new_plan_json_str)#

Add a new MAPF plan to the ADG.

This function takes a JSON string representing a new MAPF plan and adds it to the ADG (Action Decision Graph). If necessary, it utilizes the backup planner to ensure the plan is integrated correctly.

Parameters:

new_plan_json_str – A JSON string with a new MAPF plan with the following schema: 

{
  "success": true,
  "plan": [ [(row, col, timestep, task_id), ... ], ... ],
  "congested": false,
  "stats": { ... }
}

  • plan: a list of lists of tuples, where each outer list corresponds to a single robot’s path. Each inner list contains tuples of the form (row, col, timestep, task_id):

    • row: The row index of the robot’s position on the grid.

    • col: The column index of the robot’s position on the grid.

    • timestep: The timestep at which the robot should be at the specified position.

    • task_id: The task identifier, or -1 if no task is associated.

  • success: indicates whether the planner successfully produced collision-free paths for all robots. If false, the backup planner is invoked.

  • congested [Optional]: indicates whether congestion was detected, typically defined as more than half of the robots making no progress.

  • stats [Optional]: contains additional planner statistics recorded by LSMART.

bool isInitialized()#

Check if the system has been initialized.

Returns:

a boolean indicating whether the system has been initialized.

bool invokePlanner()#

Determine whether to invoke the planner.

Returns:

a boolean indicating whether the planner should be invoked. In practice, the MAPF planner can have a while loop checking for this condition to decide whether to request a new MAPF instance from the system.