A library of various mapping function for habitathttps://github.com/HongyanZhi/Habitat-Mapping-Implement.git platform
TopDownMap Coordinate System : The coordinate center maybe locates somewhere in the world
OccupiedMap Coordinate System : The coordinate center locates on the left-top of the image
OccupiedMap : project the depth image into the top down map
SemanticOccupiedMap : project the semantic image into the top down map which each pixels has some semantic meanings
SceneObjectList : Object average coordinate based on current observation
- Depth image as input must be in the scale of 10 meters, which is the default setting of DEPEH_SENSOR
- Please carefully check if the semantic images corresponding to the rgb images
- This library is suitable both for discrete and continuous enviroment; Please update the agent location by position in discrete environment and update the agent location by position or action in continuous environment
Please run the demo functions in example.py and view the reference visualization in folder visualization. Example test on v0.2.1 both for habitat-lab and habitat-sim
- Init
app = Application((image_width, image_height), fov, depth_threshold=1, resolution_meter2pixel=0.005, occupied_map_size, camera_height, free_index=1, occupied_index=2)- Api
# parse occupied map
occupied_map = app.parse_depth_topdownmap(depth_img)
# parse semantic occupied map
semantic_map = app.parse_semantic_topdownmap(depth_img, semantic_img)
# update agent location by coordinate
position = env._sim.get_agent_state().position.tolist()
rotation = env._sim.get_agent_state().rotation
app.update_pos2map_by_cooardinate(position, rotation)
# update agent location by action
app.update_pos2map_by_action(forward_step2tenmeter, turn_angle2degree, action)
# stack new map to old map
occupied_map = app.update_occupied_map(new_occupied_map, old_occupied_map)
semantic_map = app.update_semantic_map(new_semantic_map, old_semantic_map)