Maps & Occupancy Grids¶

hivebr supports two environment representations at the same time:

Analytic circular obstacles (world.obstacles) — fast, good for prototyping and for adding obstacles interactively.
Image-based occupancy grids (world.grid_map) — pixel maps converted to binary/inflated grids for collision and LiDAR.

Occupancy grids¶

Point the config to an image and give the resolution in metres per pixel:

world:
  map_image: "maze.png"
  resolution: 0.1
  show_inflation: true

Dark pixels (< 128) are treated as occupied. The grid is inflated by the robot radius (plus a 5 cm safety margin) so a point robot never cuts corners.

Loading a map at runtime¶

sim.world.load_map("path/to/map.png", resolution=0.1)
sim.grid_map = sim.world.grid_map

The Load Map button in the GUI does the same and asks for the resolution.

Building a grid programmatically¶

from hivebr.engine import Simulator

grid = my_boolean_array                  # True = occupied
sim.set_map_from_grid(grid, resolution=0.1, name="generated_map.png")

Collision and ray casting¶

occupied = sim.world.grid_map.is_occupied(x, y, use_inflated=True)
distance, hit = sim.world.grid_map.cast_ray(x, y, angle, max_range)

Generating benchmark maps¶

MapGenerator builds classic benchmarking patterns.

from hivebr.engine.map_generator import MapGenerator

grid = MapGenerator.create_grid(type="u")      # "u", "l", "random"
MapGenerator.save(grid, "hivebr/engine/assets/mapa_u.png")

Patterns available:

`type`	Description
`u`	U-shaped corridor.
`l`	L-shaped corridor.
`random`	Randomised maze from a seed.

See MapGenerator API and OccupancyGridMap API.

Optional dependencies

Efficient inflation uses scipy.ndimage.binary_dilation, and image loading uses Pillow. Install them if you rely on occupancy grids.