Unitree G1
Bipedal humanoid robot by Unitree Robotics, designed as a compact research, development, and developer platform.
Applications
- Research
- Home Assistance
**π0 (pi-zero)** is Physical Intelligence's flagship Vision-Language-Action (VLA) model, announced in October 2024 and described in the paper 'π0: A Vision-Language-Action Flow Model for General Robot Control' (arXiv:2410.24164, authors: Black, Brown, Driess et al.). π0 is the first publicly known 'generalist robot policy' that performs dexterous, long-horizon (multi-minute) manipulation sequences across many different robots — including single- and dual-arm Franka manipulators, ALOHA, UR5, and the Mobile ALOHA humanoid.
The architecture consists of three layers: (1) **VLM backbone** — PaliGemma 3B (Google) as the vision-language foundation, augmented with additional action tokens. (2) **Action expert** — a transformer-style network of ~300M parameters generating a stream of continuous actions (joint velocities) via **flow matching** (instead of action discretization as in RT-2 / OpenVLA — flow matching enables inference in full continuous precision and reduces noise). (3) **Action chunking** — the model predicts 50 future actions simultaneously (a 1-second chunk at 50 Hz), substantially reducing effective latency and producing smooth, coordinated movements.
Training data: ~10,000 hours of demonstrations across **8 robotic platforms** collected by Physical Intelligence + the Open X-Embodiment dataset (1M trajectories). Demonstrations come from teleoperation (VR rig + ALOHA). Tasks in the eval suite: folding T-shirts, packing Amazon-style shipping boxes, clearing a table, loading a dishwasher, folding a cardboard box. **π0-base** (February 2025) was released as open weights under Apache 2.0 (the first 'generalist' weights in history).
Inference: ~1 H100 GPU for full π0 (3B parameters), latency ~50-200 ms per action chunk (50 actions = 1 s of motion). Fine-tuning is available via the `openpi` codebase on GitHub (Apache 2.0). The **π0.5** version (April 2025) introduces co-training with additional multimodal datasets and better out-of-distribution generalization (e.g., unseen rooms).
An SDK (Software Development Kit) is a curated set of libraries, interfaces, tools, sample code, and documentation intended for building applications and integrating with a specific hardware device, platform, or service. In robotics, an SDK typically exposes device control, telemetry, sensor access, configuration, and execution functions, significantly reducing the time-to-first-integration for developers targeting a specific robot or platform.
A Runtime is the environment or execution layer used to run code, load libraries, manage dependencies, and operate applications or services — either in real time or during normal system operation. In robotics this includes real-time operating system (RTOS) runtimes, ROS 2 executor runtimes, containerised execution environments (Docker, podman), and embedded C++ runtimes on microcontrollers.
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.).
Demonstrations inside Physical Intelligence: at-scale T-shirt folding (50+ shirts in a row on single- and dual-arm ALOHA), packing shipping boxes, clearing a table, loading a dishwasher, folding a cardboard box. Early partner deployments (NDA — non-public). The π0-base open-weights version is used by ~200 academic teams for fine-tuning in the first 4 months after release.
github.com/Physical-Intelligence/openpi ~4.2k★, ~480 forks (4 months after release). HuggingFace `physical-intelligence/pi0-base` ~25k downloads/month. Active community on X/Twitter (@physical_int ~45k followers). Growing community of fine-tunes on specific task domains.
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.
Figure 03 is the third-generation humanoid robot from Figure AI, designed for Helix, home environments, and scalable mass production.
Compact, high-dynamics bipedal humanoid by MagicLab. 140 cm, 40 kg, 24–50 DOF, walking speed up to 2.5 m/s. Unveiled on 8 July 2025 with martial arts and acrobatic demos.
Tutor Intelligence semi-humanoid bimanual robot designed for complex manual factory and warehouse work; a fleet of 100 units operates in Data Factory 1 generating data for foundation-model training.
Ubuntu 24.04 LTS 'Noble Numbat' — supported until April 2029. The host for ROS 2 Jazzy.
Open weights for π0-base (Apache 2.0) available since February 2025. The full training database is closed — fine-tuning is only possible on your own datasets. The `openpi` code is on GitHub.
License family: Permissive
Variant optimized for inference on Jetson AGX Orin — 4-bit quantization, ~30 ms per action chunk.
Second-generation co-trained on multimodal datasets — better out-of-distribution generalization (e.g., new rooms, unseen objects).
Release of π0-base weights under Apache 2.0 along with the `openpi` codebase for fine-tuning.
arXiv:2410.24164 preprint published — first public unveiling of π0 with demonstrations and benchmarks.
