Unitree G1
Bipedal humanoid robot by Unitree Robotics, designed as a compact research, development, and developer platform.
Applications
- Research
- Home Assistance
ORB-SLAM3 is the third generation of the flagship Visual SLAM framework developed since 2014 at the Robotics, Perception and Real-Time Group (RoPeRT) at the University of Zaragoza, led by Prof. Juan D. Tardós, José M. M. Montiel, and Carlos Campos. The first ORB-SLAM version introduced feature-based SLAM based on ORB (Oriented FAST and Rotated BRIEF) descriptors. ORB-SLAM2 (2017) added stereo and RGB-D. ORB-SLAM3 (2021) introduces Visual-Inertial SLAM, multi-map tracking, and full fish-eye support. GPLv3 license.
ORB-SLAM3 architecture covers four threads: (1) tracking — estimating camera pose for every frame, (2) local mapping — maintaining and optimizing the local map, (3) loop closing — loop detection via DBoW2 bag-of-words and Sim(3) optimization, (4) full BA — full bundle adjustment for the entire map. Also built in is the Atlas — a multi-map management system with the ability to merge maps.
Key components: g2o (graph optimization), DBoW2 (visual place recognition), OpenCV (feature extraction), Eigen (linear algebra). Performance: ~30 Hz on CPU for a 640×480 image on a PC; ~10 Hz on Jetson Orin Nano. Limitations: no dense map (only sparse ORB 3D points), no semantics, vulnerable in texture-poor environments. A ubiquitous benchmark for every new Visual SLAM publication.
A Perception Stack encompasses the software layers that process data from cameras, LiDARs, IMUs, microphones, and other sensors in order to recognise the surrounding environment, perform localisation, detect and track objects, and interpret the scene. It is typically the first processing stage in an autonomous robot's data pipeline, feeding its outputs to planning and control stacks.
An API Library is a software package that exposes programmatic interfaces for communicating with a device, service, or system. In robotics it typically forms a lightweight integration layer built on top of the manufacturer's official API or an open-source project, abstracting low-level protocol details and providing language-native bindings (Python, C++, Java, etc.).
A family of open-source robotic perception libraries: Visual SLAM (ORB-SLAM, RTAB-Map), computer vision (OpenCV), point cloud processing (PCL, Open3D), detection and segmentation (YOLO, Detectron2).
Reference benchmark in EuRoC MAV (drones), TUM RGB-D (manipulators), KITTI Visual Odometry (autonomous cars), TUM VI (Visual-Inertial). Used in ICRA / IROS / RSS papers as a baseline. Production implementations: SkyDio drones, Magic Leap AR research, NASA JPL Mars helicopter Ingenuity (derivatives).
github.com/UZ-SLAMLab/ORB_SLAM3 ~7.5k★, ~3k forks, > 2300 publications citing ORB-SLAM3 (Campos et al., IEEE TRO 2021). Active ResearchGate group.
Bipedal humanoid robot by Unitree Robotics, designed as a compact research, development, and developer platform.
Unitree H1 is a full-size general-purpose humanoid robot (~180 cm, ~47 kg). Bipedal, 5 DOF per leg + 4 DOF per arm, 3.3 m/s walking speed, 360° perception via 3D LiDAR + depth camera, Unitree M107 PMSM joint motors with ~360 N·m peak knee torque. Standard compute: Intel Core i5/i7; optional NVIDIA Jetson Orin NX.
Boston Dynamics bipedal humanoid robot. The fully electric generation unveiled in 2024 succeeds the hydraulic Atlas that was retired after more than a decade of research.
Requires OpenCV ≥ 4.4, Pangolin (viewer), Eigen3, g2o, DBoW2. Ubuntu 22.04 + ROS 2 Humble recommended.
License family: Strong Copyleft
ROS 2 Humble support, build fixes on Ubuntu 22.
Minor fixes, better Eigen 3.4 compatibility.
Stable production release; Atlas multi-map.
First public beta with Visual-Inertial SLAM.
Added stereo and RGB-D; full DBoW2 integration.
First version of monocular Feature-SLAM (Mur-Artal et al., IEEE TRO).
