Path Planners

Planners compute a geometric path from a start pose to a goal. They live in examples/planners and share the BasePlanner interface.

BasePlanner

class BasePlanner(ABC):
    def __init__(self, bounds, robot_radius=0.3, seed=None): ...
    def plan(self, start, goal, obstacles, occupancy_map=None) -> (path, DebugData): ...
    @abstractmethod
    def _plan(self, start, goal, obs_list, occupancy_map=None): ...

• bounds is (xmin, xmax, ymin, ymax) in metres.
• obstacles is world.obstacles (the engine's [x, y, r] arrays).
• occupancy_map is an optional OccupancyGridMap.
• plan returns a list of (x, y) from start to goal (or None) and a DebugData object describing the search tree/roadmap for visualization.

Subclasses only implement _plan. Shared collision helpers (_point_free, _segment_free, _in_bounds, _random_sample, _steer, …) are provided by the base class.

Available planners

Planner	Import	Class / type
A*	examples.planners.astar	AStarPlanner — grid-based A* on an inflated occupancy grid.
RRT	examples.planners.planners	RRTPlanner — rapidly-exploring random tree.
RRT*	examples.planners.planners	RRTStarPlanner — asymptotically optimal RRT.
PRM	examples.planners.planners	PRMPlanner — probabilistic roadmap.
BIT*	examples.planners.planners	BITStarPlanner — batch informed trees.
OMPL	examples.planners.ompl_planner	OMPLPlanner / OMPLPlannerAdapter — wraps an OMPL planner (RRT, RRTConnect, PRM, BIT*, …). Requires ompl.
RRT-Dubins	examples.planners.rrtdubins	Dubins-path variants for car-like (Ackermann) B-bots.

Connecting a planner to the engine

PlannerController turns any planner into an engine controller and draws the path plus debug structure on screen.

from examples.planners.planners import RRTStarPlanner
from examples.controllers.planner_controller import PlannerController
from hivebr.engine import Simulator, load_config

config = load_config("config/default.yaml")
bounds = (0, config["world"]["width"], 0, config["world"]["height"])

planner = RRTStarPlanner(bounds, robot_radius=config["robot"]["radius"], seed=1)
controller = PlannerController(planner)

sim = Simulator(config=config, controller=controller.controller,
                draw_callback=controller.draw_callback)
sim.run()

Writing your own planner

from examples.planners.base_planner import BasePlanner, DebugData

class StraightPlanner(BasePlanner):
    def _plan(self, start, goal, obs_list, occupancy_map=None):
        path = [start, goal]
        return path, DebugData()

See also the Controllers page for alternative (reactive) navigation strategies.