GeezerVison

Face Detection, Recognition, Gaze Estimation, and Visual Speech Activity for Human-Robot Interaction

Base Docker setup for ROS 2 Vulcanexus (Jazzy) + PyTorch. This repo is intentionally minimal: a single docker/ folder that builds a reproducible image to bootstrap the multimodal perception pipeline (sound + vision + people/face subsystems).

A set of HRI-related ROS message, conformant with the ROS4HRI specification

Orbbec Astra / Dabai DC1 driver for the Mercury X1 robot (ROS 2 Galactic). Branch mercury-galactic: color stream and depth/color sync turned on in the launch files, best-effort QoS for image publishers, and RGB QoS settings passed through from the launch files.

image_pipeline for the Mercury X1 robot (ROS 2 Galactic), based on the galactic branch. Branch mercury-galactic: the depth_image_proc RegisterNode accepts NodeOptions, so it can be configured when used as a composable node.

ackermann_msgs for the Mercury X1 robot (ROS 2 Galactic), based on 2.0.2. Branch mercury-galactic: build fixes for Galactic (C99 default in CMake) and cleanup of AckermannDrive.msg comments. Message fields are unchanged.

usb_cam for the Mercury X1 robot (ROS 2 Galactic), based on 0.4.2. Branch mercury-galactic: parameters can be changed at runtime (the camera restarts with the new settings), a bottom_camera calibration, and the launch file renamed to usb_cam_launch.py.

Nav2 for the Mercury X1 robot (ROS 2 Galactic), based on 1.0.12. Branch mercury-galactic: map_server republishes the map periodically, Foxy-style map_saver (fixes saving on Galactic), lifecycle_manager ignores duplicate startup requests, and a tuned RViz config.

BehaviorTree.CPP utilities to work with ROS2

Integration packages to enable communication between ROS 2 and VDA5050 protocol, in particular for those AMRs with a Nav2 stack.