Unitree Go2 is a compact quadruped (“robot dog”) developed by Unitree Robotics as the successor to the Go1. Targeted at education, research, and early-stage field applications, Go2 combines high-torque electric actuation with multi-sensor perception and developer-oriented software. The platform is sold in several variants (Air, Pro, X, and EDU) that share a common mechanical envelope and support add-ons such as 4D LiDAR and extended-capacity batteries.
Unitree Go2 Quadruped Robot
Form factor and mechanics
Go2 maintains a small, lab-friendly footprint with a standing size of ~70 × 31 × 40 cm and an all-up weight of ~15 kg including the battery. The chassis combines aluminum alloy and high-strength engineering plastics, balancing stiffness with impact tolerance for classroom and indoor testing.
Perception: 4D LiDAR and vision
A defining upgrade over earlier models is support for Unitree’s self-developed 4D LiDAR L1, offering a 360° × 90° hemispherical field with a minimum detection distance of ~0.05 m, designed to reduce blind spots and bolster near-field obstacle detection for indoor navigation and tight maneuvers. The LiDAR complements onboard camera sensing and can be paired with app-level control and streaming.
Power system and endurance
Go2 increases standard battery capacity to 8,000 mAh, with an optional 15,000 mAh “ultra-long life” pack. Unitree notes a higher system voltage of ~28.8 V (28–33.6 V operating window) to improve actuator efficiency and control stability. Third-party listings specify the 8,000 mAh pack at ~236.8 Wh, with a battery mass of ~2 kg. Typical endurance ranges from ~1–2 hours depending on usage and configuration.
Mobility and handling
Depending on the version, Go2 advertises walking/running speeds up to ~3.7 m/s (with higher “limit” speeds specified for certain configurations), payloads around ~7–8 kg, and basic step-climb capabilities suitable for indoor obstacles and stairs. Cable routing and thermal design changes improve serviceability and thermal performance at the joints compared with earlier consumer models.
Software and secondary development
Unitree positions Go2 as a secondary-development platform, providing developer documentation and model variants that support higher-compute options (e.g., Jetson-class modules) for SLAM, imitation learning, and embodied AI experiments. Documentation and product lists for the Go2 family (including dimensions and weights) are maintained in Unitree’s developer portal.
Technology and Specifications
Mechanical and electrical (representative)
-
Standing dimensions: ~70 × 31 × 40 cm
-
Lying/crouching dimensions: ~76 × 31 × 20 cm
-
Weight (with battery): ~15 kg
-
Materials: aluminum alloy + high-strength engineering plastic
-
Operating voltage: ~28–33.6 V (noted 28.8 V nominal for efficiency)
Performance (by variant)
-
Payload: ~7–8 kg (higher “limit” values vary by model)
-
Speed: up to ~3.5–3.7 m/s (with higher “limit” speeds cited in marketing materials)
-
Climb/obstacle ability: indoor steps and moderate slopes (values vary by version; EDU materials cite ~16 cm step height and up to ~40° ramps).
Energy and runtime
-
Standard battery: 8,000 mAh; optional: 15,000 mAh
-
Approx. endurance: ~1–2 hours (usage-dependent)
-
Battery spec (3rd-party listing): 8,000 mAh ≈ 236.8 Wh; ~2 kg pack mass.
Sensors and options
-
4D LiDAR L1: 360° × 90° FOV; min detection ~0.05 m
-
Cameras: near-field navigation and app-level video streaming; configurations vary by trim and region.
Note: Detailed specifications differ by Air, Pro, X, and EDU variants and may change with firmware and production updates; consult current datasheets before procurement.
Applications and Use Cases
Education and teaching labs
Go2’s small footprint and price point make it a fit for robotics courses covering locomotion, state estimation, and reinforcement/imitative learning. Fast setup, app control, and readily available spares (batteries, chargers, controllers) help maintain high lab throughput.
Research and prototyping
Research groups use Go2 for SLAM, autonomous navigation in cluttered indoor spaces, and human-robot interaction (HRI) studies. The optional 4D LiDAR expands dense mapping, while higher compute options support dataset logging and model deployment.
Facilities inspection and training
While not as rugged as Unitree’s industrial models, Go2 serves as a testbed for inspection workflows in controlled environments—e.g., mock factories, data centers, or labs—where navigating narrow corridors and steps is important. EDU-oriented bundles sometimes add third-party LiDAR (e.g., MID-360) for research flexibility.
Outreach, demos, and events
Go2’s app-based control and compact size suit public demonstrations and tech outreach, where safe interaction, repeatable routines, and moderate runtime are priorities. Retailer materials emphasize remote control, 3D mapping, and ISS 2.0 positioning for smoother demos.
Advantages / Benefits
-
Accessible legged mobility: Brings dynamic quadruped locomotion to classrooms and small labs at a comparatively lower cost than industrial units, widening participation in legged-robotics research.
-
Enhanced near-field perception: The 4D LiDAR’s hemispherical coverage and short minimum range reduce blind spots for tight-space navigation and obstacle avoidance.
-
Improved energy architecture: Higher battery capacity and elevated system voltage (to ~28.8 V) support motor efficiency, stability, and practical runtimes for coursework and experiments.
-
Variant flexibility: Air/Pro/X/EDU trims and third-party LiDAR options allow teams to align budget with sensing and compute needs.
Comparisons (if relevant)
-
Go2 vs. Go1: Go2 adds a larger battery option, higher system voltage, and standardized support for 4D LiDAR L1 with 360° × 90° coverage. Go1 prioritized camera-centric perception and the original Intelligent Side-Follow feature set. (Exact feature differences depend on sub-variants).
-
Go2 vs. industrial quadrupeds (e.g., B1/B2): Go2 focuses on education and research with lower mass and indoor-friendly packaging, while B-series platforms emphasize payload, ingress protection, and outdoor duty cycles. Buyers evaluating inspection in harsh environments often graduate to industrial lines.
Pricing and Availability
Unitree’s official shop lists Go2 with a starting price of ~US$2,800 (region and variant dependent), with options for Air, Pro, X, and EDU. Accessories such as batteries, controllers, chargers, and joint motors are sold separately on official storefronts. Third-party resellers publish EDU bundle pricing—e.g., Go2 EDU PLUS with MID-360 LiDAR cited at around US$13,205 (retailer listing; subject to change). Shipping charges, taxes, and lead times vary by location and stock. Prospective buyers should confirm current pricing and configurations with sales.
FAQ
What is Unitree Go2?
Unitree Go2 is a small quadruped robot for education and research, offering multi-sensor perception and developer-oriented options across Air, Pro, X, and EDU variants.
How does Unitree Go2 work?
Go2 uses high-torque electric joints and real-time control to execute stable gaits, while cameras and optional 4D LiDAR L1 provide 3D awareness for obstacle avoidance, teleoperation, and mapping.
Why is Unitree Go2 important?
It lowers the barrier to legged-robotics research and education, providing a standardized platform for SLAM, embodied AI, and HRI coursework without the cost and complexity of industrial units.
What are the benefits of the 4D LiDAR option?
The L1 module expands the field of view to 360° × 90° with short-range detection down to ~0.05 m, reducing blind spots and improving navigation in tight indoor spaces.
Summary
The Unitree Go2 advances accessible legged mobility with a standardized 70-cm class frame, expanded battery options, and a 4D LiDAR upgrade that improves near-field perception. Its variant strategy (Air/Pro/X/EDU) and developer resources make it a versatile platform for coursework, SLAM and navigation research, and demonstrations. For teams seeking practical hands-on experience with quadruped autonomy at a modest footprint and cost, Go2 represents a well-documented, widely adopted starting point.