Vision-based localization using ArUco markers

This package implements a localization systtem using ArUco markers.

Markers may be fixed or roaming. Cameras view the markers and calculate roaming marker poses relative to fixed markers. Using fixed marker poses relative to world the roaming markers world pose is known.

Each camera must see at least 1 fixed marker.

Multiple cameras can be used, and pose information is published over ZMQ by each camera node. A server subscribes to all the pose information and makes it available over a HTTP API.

Installation

The same package can be installed on camera nodes and server.

pip install -e .

Usage

A configuration file needs to be provided to camera nodes and server. An example is provided in /config/config.yaml.

Once installed the package provides 2 entrypoints, one for camera node and one for server.

To run on camera node:

localization_node --log <log-level> <config_file> <camera_name>

To run on localization server:

localization_server --log <log_level> <config_file>

Camera Calibration

To obtain camera intrinsics to populate the config file, use the provided calibration tool. The calibration process requires the use of a ChArUco board.

Run the calibration tool on the camera node:

camera_calibrate -v

Use the -h flag for more information, including how to set ChArUco board parameters.

You can view the camera view to help in board alignment by directing a browser to http://<camera node host>:5127.