Go2

Source-backed capabilities

• Capability areas: Autonomous navigation
• Workflows: Industrial inspection
• Operating environments: Research lab
• Target markets: Robotics researchers, developers and inspection teams seeking an accessible quadruped platform.
• Deployment model: Commercial product sold as a development and light-operations platform.
• Multi-terrain locomotion

  Demonstrated - 2026-10-06 - The controller used body tilt and ground-reaction information to maintain posture across uneven surfaces.

  Source
• Navigation and obstacle avoidance

  Demonstrated - 2026-09-08 - Navigate suspended construction scaffolding with Akari learned locomotion and LiDAR obstacle detection. - environment: Suspended construction scaffolding

  Source
• Outdoor mobility assistance

  Research demonstration - 2026-07-21 - Guided a handler through real indoor and outdoor obstacle courses while avoiding obstacles and pedestrians. - environment: real indoor and outdoor guide-dog obstacle courses

  Source
• Multi-terrain locomotion

  Research demonstration - 2026-06-14 - The paper reports robust zero-shot deployment on a physical Unitree Go2 during rough-terrain traversal.

  Source
• Navigation and obstacle avoidance

  Research demonstration - 2026-06-09 - AgniNav standardizes cross-embodiment monocular local navigation through a configurable collision-envelope representation.

  Source
• Stair climbing

  Research demonstration - 2026-02-16 - The paper studies a two-stage reinforcement-learning approach for improving quadruped performance on U-shaped stair climbing.

  Source
• Stair climbing

  Research demonstration - Climbed hollow stairs with inclines up to 55 degrees. - environment: stairs up to a 55-degree incline

  Source
• Human following

  Claimed - Unitree presents Go2 with intelligent obstacle detection and avoidance plus a side-following mode.

  Source
• Whole-body loco-manipulation

  Research demonstration - 2026-03-02 - Safe whole-body loco-manipulation paper validates model-learning control on Unitree Go2

  Source
• Whole-body loco-manipulation

  Research demonstration - 2025-07-27 - The paper proposes a risk-adaptive reinforcement-learning framework that enables a quadruped to walk on its hind legs for loco-manipulation.

  Source
• Whole-body loco-manipulation

  Claimed - 2026-07-07 - Grasp objects at ground level and manipulate with both hands while all four feet remain planted. - environment: simulated bimanual quadruped manipulation tasks

  Source
• Vision-language navigation

  Research demonstration - 2026-03-06 - Followed natural-language navigation instructions on the physical quadruped in real-world environments. - environment: real-world instruction-following navigation courses

  Source
• Load carrying

  Research demonstration - 2025-07-27 - Real-world deployment on a Unitree Go2 robot demonstrates cart pushing, obstacle probing, payload transport, and robustness to disturbances.

  Source

Exact-product configurations are shown without inferring supply, purchase, customer status, standard fit, deployment or partnership.

• Akari learned locomotion controller

  Evaluation - Controlled By - Physical - A Unitree Go2 with Akari locomotion software navigated suspended construction scaffolding and freed legs caught on sheets or cables during an Oriental Shiraishi field test.

  Source

• LocoPilot: Single-Operator Bilateral Teleoperation for Loco-Manipulation Policy Learning

  2026-10-20 - Research Publication - LocoPilot lets one operator coordinate a Unitree Go 2 quadruped and AgileX Piper L arm through bilateral teleoperation. A unified 11D action interface connects demonstration collection and policy execution. The IIT team fine-tunes MolmoAct 2 on pick-and-place demonstrations and tests it on the physical system.

  Source
• CusADi2: Latency-Optimized Batch Computation of Robot Kinematics, Jacobians, and Hessians on the GPU

  2026-10-18 - Research Publication - CusADi2 reduces batch-computation latency by parallelizing both across and within robot-model evaluations on GPUs. The authors benchmark kinematics, Jacobians and Hessians across67 MuJoCo Menagerie robots and report lower latency at batch sizes64–4096 than evaluated alternatives.

  Source
• CQDC-Warp: GPU-Accelerated Differentiable Quasi-Dynamic Simulation for Contact-Rich Manipulation

  2026-10-15 - Research Publication - CQDC-Warp uses GPU-batched differentiable quasi-dynamic simulation to plan contact-rich trajectories. Experiments cover dexterous-hand and whole-body models, with a limited physical replay on a Sharpa hand.

  Source
• Load-Aware Quadruped Navigation with Online Contact-based Geometry Estimation

  2026-10-10 - Research Publication - LOADNav learns navigation that accounts for carried-object geometry; CoGE estimates unknown geometry from contact events. The study evaluates policies in simulation and on a Unitree Go2 carrying elongated payloads through clutter.

  Source
• Milo, a Fully Autonomous Indoor/Outdoor Robotic Guide Dog

  2026-07-21 - Research Publication - A modified Go2 guided a handler through real indoor and outdoor obstacle courses with onboard obstacle and pedestrian avoidance.

  Source
• Beyond Transformers: Linear Attention Policy for Open-Vocabulary Object Goal Navigation

  2026-07-21 - Research Publication - A configured Go2 EDU found language-specified indoor objects in 41 of 50 physical trials.

  Source
• Predictive Training with Latent Imagination for Visual Quadruped Navigation

  2026-07-20 - Research Publication - A physical Go2 avoided indoor clutter and an approaching person outdoors after zero-shot transfer.

  Source
• Learning Agile Navigation in Crowded Environments for Quadruped Robots

  2026-07-16 - Research Publication - A sensor-configured Go2 navigated complex indoor and outdoor scenes with moving obstacles.

  Source
• NavCMPO: Critic-Guided MeanFlow Policy Optimization for Adaptive Navigation

  2026-07-16 - Research Publication - A D435i-equipped Go2 navigated cluttered corridors, outdoor paths and complex indoor obstacles.

  Source
• From Language to Navigation Goals: A Vision-Language Approach for Semantic Navigation of Mobile Robots Using RGB-D Perception

  2026-07-15 - Research Publication - A RealSense-equipped Go2 validates language-conditioned target perception; full navigation remains a simulation result.

  Source
• Stop to Decide: Latency-Aware Proprioceptive Navigation Primitives for Mapping-Free Quadruped Inspection

  2026-07-13 - Research Publication - A Go2 EDU completed line, corridor and stair sequences in 18 of 20 physical course trials.

  Source
• TAC-LOCO: Unified Whole-Body Control for Quadrupedal Tactile-Informed Loco-Manipulation

  2026-07-11 - Research Publication - A configured Go2 walked while pulling grasped objects and adapting grip force to changing loads.

  Source
• Calf-Integrated Arms for Bimanual Quadruped Loco-Manipulation

  2026-07-07 - Research Publication - Most quadruped loco-manipulation designs trade manipulation capability against stance. A trunk-mounted arm sits high and usually carries a single arm; using the legs as manipulators lifts the manipulating leg off the ground; and even leg-mounted grippers reach two-handed tasks only by rearing onto the hind legs. This paper integrates a manipulator with a prismatic slider, two revolute joints, and a gripper into each

  Source
• TypeGo: An OS Runtime for Embodied Agents

  2026-07-06 - Research Publication - Large language models (LLMs) can plan behavior for embodied agents from natural language, but treating the LLM as a request/response oracle on the critical path is fundamentally at odds with real-time control and concurrent goals. We argue for an operating-system-style runtime for embodied agents, and instantiate this idea in an early prototype, TypeGo. TypeGo structures LLM-based planning as asynchronous loops at mu

  Source
• CacheMPC: Certified Cached Model Predictive Control for Quadruped Locomotion

  2026-06-26 - Research Publication - Model Predictive Control (MPC) is the standard predictive layer in hierarchical quadruped controllers, but the per-cycle QP solve limits the update rate achievable on embedded processors.

  Source
• Unleashing Infinite Motion: Scaling Expressive Quadrupedal Motion via Generative Video Priors

  2026-06-26 - Research Publication - Quadruped robots have achieved remarkable locomotion, yet their behavioral repertoire remains confined to a few gaits--far from the expressive, companion-like presence long envisioned for them.

  Source
• StairMaster: Learning to Conquer Risky Hollow Stairs for Agile Quadrupedal Robots

  2026-06-24 - Research Publication - Climbing hollow stairs remains a challenging problem for quadruped robots due to the high risk of leg trapping, severe depth sparsity, and high-frequency depth-sensing noise.

  Source
• RoboAtlas: Contextual Active SLAM

  2026-06-24 - Research Publication - We present RoboAtlas, a contextual Active SLAM framework that adaptively balances geometric exploration and semantic reasoning using a scalable 3D semantic mapping system, OpenRoboVox.

  Source
• MPC-Injection: Biasing Off-Policy Locomotion RL Toward Controller-Induced Behavior Basins

  2026-06-24 - Research Publication - Reinforcement learning (RL) for locomotion frequently converges to locally optimal but undeployable behaviors, such as vibrating limbs or scooting on the torso, that maximize return without producing a usable gait. We present MPC-Injection, a low-overhead method that steers RL toward a designer-preferred gait by inserting transitions into the replay buffer from a model predictive controller solving the same Markov de

  Source
• fARfetch: Enabling Collocated AR-HRC in Large Visually Diverse Environments with VLM-Driven AR Content Adaptation

  2026-06-23 - Research Publication - Augmented Reality (AR) can improve collocated human-robot collaboration by making robot state and intent visible and enabling intuitive control, yet large, visually diverse environments like the outdoors challenge both interaction and content legibility, especially at long distances and beyond visual line of sight.

  Source
• ForEnt: A Multi-Modal Dataset for Characterizing Quadruped Robot Entrapments in Forest Environments

  2026-06-18 - Research Publication - Legged robots are increasingly deployed in forests for ecological surveying and monitoring, yet their autonomy is often interrupted consequent to the challenges posed in traversing forest environments.

  Source
• SemGeoNav:A Safety-Guided Visual Navigation Approach with Semantic Reasoning and Geometric Planning

  2026-06-15 - Research Publication - Learning-based visual navigation has enhanced semantic goal-reaching capabilities.

  Source
• FlashNav: Ultra-Fast Policy Training for Robot Navigation within 20 Seconds

  2026-06-14 - Research Publication - Deep reinforcement learning has shown strong potential for robot navigation, but its practical deployment is still limited by the long wall-clock cost of policy training.

  Source
• LoComposition: Terrain-Adaptive Energy-Efficient Quadruped Locomotion without Gait Priors

  2026-06-14 - Research Publication - Learning-based quadrupedal locomotion typically relies on complex reward formulations that entangle task specification, operational limits, gait preference, and terrain adaptation within a single optimization objective.

  Source
• Kine2Go: Kinematic dataset for the Unitree Go2 robot with diverse gaits and motions

  2026-06-12 - Research Publication - The recent popularity of robotics, combined with the steadily decreasing cost of robotic hardware, has lowered the entry barrier to robotics research and enabled rapid advancements in the field. One of the primary examples is the Unitree Go2 quadruped robot, which is often used by researchers in the areas of locomotion, navigation, control, and others. Many researchers use the Go2 robot in combination with techniques

  Source
• nomagiclab/kine2go-pipeline

  2026-06-12 - Research Publication - The complete end-to-end pipeline for quadrupedal motion imitation: cross-morphology retargeting, imitation policy training, and dataset generation for the Unitree Go2 robot.

  Source
• AgniNav: Configuration-Driven Cross-Embodiment Local Planning for Robot Navigation

  2026-06-09 - Research Publication - Monocular local navigation is attractive for lightweight robots, but existing vision-based policies often couple perception to a specific body, camera height, and footprint, making transfer from wheeled bases to legged platforms dependent on retraining or active depth hardware.

  Source
• Bayesian Optimization for Learning Nonlinear MPC in Autonomous Agent Navigation

  2026-06-08 - Research Publication - Real-time autonomous navigation in dynamic, unknown environments remains a fundamental challenge for mobile robotics.

  Source
• Continual Quadruped Robots Coordination via Semantic Skill Discovery

  2026-06-06 - Research Publication - Multi-quadruped coordination has attracted increasing attention due to its enhanced payload capacity, broader contact coverage, and improved adaptability to challenging tasks.

  Source
• Autonomous Navigation System for Library Service Robot Based on Unitree Go2 Edu

  2026-06-02 - Research Publication - Libraries require autonomous robots to move quietly through narrow aisles while remaining safe around readers, chairs, bags, and carts.

  Source
• PlatonicNav: Unveiling Semantic Correspondence in Navigation with Platonic Topological Maps

  2026-06-01 - Research Publication - Embodied visual navigation, where an agent perceives a complex environment and acts to reach a goal from raw sensory input, underpins a wide range of applications such as household service robotics, assistive robotics, and large-scale autonomous exploration.

  Source
• V4C38/spectacles-dimensional-os

  2026-05-30 - Research Publication - Snap Spectacles AR integration with Dimensional OS - contoller for Unitree GO2 and G1 robots.

  Source
• High-Load-Density Electro-Permanent Magnetic Foot with Controllable Adhesion for Quadruped Wall-Climbing Robots

  2026-05-29 - Research Publication - To enable reliable climbing locomotion of quadruped robots on ferromagnetic surfaces, this paper presents a high-load-density electro-permanent magnetic foot with controllable adhesion, featuring force-feedback circular Halbach-net electro-permanent magnet (CHN-EPM) adhesion units and a magnetization control system.

  Source
• TravExplorer: Cross-Floor Embodied Exploration via Traversability-Aware 3-D Planning

  2026-05-19 - Research Publication - Zero-shot Object Navigation (ZSON) has shown promise for open-vocabulary target search in unseen environments, yet most existing systems remain tied to planar representations and single-floor assumptions.

  Source
• Quadruped Parkour Learning: Sparsely Gated Mixture of Experts with Visual Input

  2026-04-21 - Research Publication - Robotic parkour provides a compelling benchmark for advancing locomotion over highly challenging terrain, including large discontinuities such as elevated steps.

  Source
• CART: Context-Aware Terrain Adaptation using Temporal Sequence Selection for Legged Robots

  2026-04-15 - Research Publication - Animals in nature combine multiple modalities, such as sight and feel, to perceive terrain and develop an understanding of how to walk on uneven terrain in an efficient manner.

  Source
• Simulator Adaptation for Sim-to-Real Learning of Legged Locomotion via Proprioceptive Distribution Matching

  2026-04-13 - Research Publication - Simulation trained legged locomotion policies often exhibit performance loss on hardware due to dynamics discrepancies between the simulator and the real world, highlighting the need for approaches that adapt the simulator itself to better match hardware behavior. Prior work typically quantify these discrepancies through precise, time-aligned matching of joint and base trajectories. This process requires motion captu

  Source
• Natural Gradient Gaussian Approximation Filter on Lie Groups for Robot State Estimation

  2026-04-11 - Research Publication - Accurate state estimation for robotic systems evolving on Lie group manifolds, such as legged robots, is a prerequisite for achieving agile control.

  Source
• PriPG-RL: Privileged Planner-Guided Reinforcement Learning for Partially Observable Systems with Anytime-Feasible MPC

  2026-04-09 - Research Publication - This paper addresses the problem of training a reinforcement learning (RL) policy under partial observability by exploiting a privileged, anytime-feasible planner agent available exclusively during training.

  Source
• Safety-Critical Centralized Nonlinear MPC for Cooperative Payload Transportation by Two Quadrupedal Robots

  2026-04-03 - Research Publication - This paper presents a safety-critical centralized nonlinear model predictive control (NMPC) framework for cooperative payload transportation by two quadrupedal robots.

  Source
• AlanD808/Go2-navigation-self

  2026-04-02 - Research Publication - It's a small project based on unitree go2 robotic dog and champ controller, aiming at accomplishing mapping and navigating without Nav2. The "3D pointcloud to 2D laserscan" converter has been created to solving the issue which 3D

  Source
• DFKI publishes Unitree Go2 quadruped state-estimation dataset

  2026-03-30 - Research Publication - DFKI Robotics Innovation Center published a Unitree Go2 dataset for legged-robot state estimation. The recordings cover indoor and outdoor Go2 locomotion with IMU and leg-odometry data captured at 1000 Hz.

  Source
• dongsheng123132/go2-openclaw-skill

  2026-03-24 - Research Publication - Control Unitree Go2 robot dog with OpenClaw AI via natural language — CycloneDDS gateway, no ROS2 needed

  Source
• darshmenon/quadruped-robotics-stack

  2026-03-22 - Research Publication - Unitree Go2 quadruped robot dog — RL locomotion training (MuJoCo + Gazebo Harmonic), ROS2 CHAMP and Quad-SDK NMPC walking controllers, PPO policy, keyboard teleop, and multi-terrain simulation

  Source
• ADMM-Based Distributed MPC with Control Barrier Functions for Safe Multi-Robot Quadrupedal Locomotion

  2026-03-19 - Research Publication - This paper proposes a fully decentralized model predictive control (MPC) framework with control barrier function (CBF) constraints for safety-critical trajectory planning in multi-robot legged systems.

  Source
• REAL: Robust Extreme Agility via Spatio-Temporal Policy Learning and Physics-Guided Filtering

  2026-03-18 - Research Publication - Extreme legged parkour demands rapid terrain assessment and precise foot placement under highly dynamic conditions.

  Source
• SLowRL: Safe Low-Rank Adaptation Reinforcement Learning for Locomotion

  2026-03-17 - Research Publication - Sim-to-real transfer of locomotion policies often leads to performance degradation due to the inevitable sim-to-real gap.

  Source
• NavThinker: Action-Conditioned World Models for Coupled Prediction and Planning in Social Navigation

  2026-03-16 - Research Publication - Social navigation requires robots to act safely in dynamic human environments.

  Source
• yusufdxb/GO2-seeing-eye-dog

  2026-03-16 - Research Publication - Assistive robotics thesis on Unitree GO2: multimodal identity-gated navigation for visually impaired users. ROS 2 Jazzy, Jetson Orin, YOLOv8, Whisper ASR.

  Source
• Rai220/go2-python

  2026-03-07 - Research Publication - Unitree Go2 robot controller via WebRTC — Python CLI and web interface

  Source
• SysNav: Multi-Level Systematic Cooperation Enables Real-World, Cross-Embodiment Object Navigation

  2026-03-06 - Research Publication - Object navigation (ObjectNav) in real-world environments is a complex problem that requires simultaneously addressing multiple challenges, including complex spatial structure, long-horizon planning and semantic understanding.

  Source
• History-Conditioned Spatio-Temporal Visual Token Pruning for Efficient Vision-Language Navigation

  2026-03-06 - Research Publication - Vision-Language Navigation (VLN) enables robots to follow natural-language instructions in visually grounded environments, serving as a key capability for embodied robotic systems. Recent Vision-Language-Action (VLA) models have demonstrated strong navigation performance, but their high computational cost introduces latency that limits real-time deployment. We propose a training-free spatio-temporal vision token prun

  Source
• Whole-Body Safe Control of Robotic Systems with Koopman Neural Dynamics

  2026-03-04 - Research Publication - Controlling robots with strongly nonlinear, high-dimensional dynamics remains challenging, as direct nonlinear optimization with safety constraints is often intractable in real time. The Koopman operator offers a way to represent nonlinear systems linearly in a lifted space, enabling the use of efficient linear control. We propose a data-driven framework that learns a Koopman embedding and operator from data, and int

  Source
• Safe Whole-Body Loco-Manipulation via Combined Model and Learning-based Control

  2026-03-02 - Research Publication - Simultaneous locomotion and manipulation enables robots to interact with their environment beyond the constraints of a fixed base.

  Source
• anabeldilab/LLM-WebRTC-Unitree-Go2

  2026-02-28 - Research Publication - Semantic LLM planner (Llama 3) to voice-control the Unitree Go2 robot generating safe, structured JSON commands. / Planificador semántico LLM para controlar el Unitree Go2 generando JSON seguros. (Note: Requires WebRTC middleware by legion1581).

  Source
• External stair-climbing paper trains and transfers policies on Unitree Go2

  2026-02-16 - Research Publication - Quadruped robots are employed in various scenarios in building construction. However, autonomous stair climbing across different indoor staircases remains a major challenge for robot dogs to complete building construction tasks.

  Source
• orisharabi/unitree-go2-follow-system

  2026-02-11 - Research Publication - Autonomous follow system for the Unitree Go2 robot using UWB tracking, YOLO-based vision and PID motion control.

  Source
• UEREBot: Learning Safe Quadrupedal Locomotion under Unstructured Environments and High-Speed Dynamic Obstacles

  2026-02-07 - Research Publication - Quadruped robots are increasingly deployed in unstructured environments. Safe locomotion in these settings requires long-horizon goal progress, passability over uneven terrain and static constraints, and collision avoidance against high-speed dynamic obstacles. A single system cannot fully satisfy all three objectives simultaneously: planning-based decisions can be too slow, while purely reactive decisions can sacrif

  Source
• Schrödinger's Navigator: Imagining an Ensemble of Futures for Zero-Shot Object Navigation

  2025-12-24 - Research Publication - Zero-shot object navigation (ZSON) requires robots to find target objects in unseen environments without task-specific fine-tuning or pre-built maps, a key capability for general-purpose service robots. Yet methods that perform well in simulation often degrade in cluttered real-world scenes with severe occlusion and latent hazards, where large unseen regions make single-scene inference brittle and unsafe. We propose

  Source
• LoGoPlanner

  2025-12-23 - Research Publication - Presents a project page for LoGoPlanner, a localization-grounded navigation framework. Combines visual-geometry-driven state estimation with planning to improve obstacle avoidance. Facilitates robust navigation and trajectory planning in environments without external localization. 🤖Can robots achieve accurate navigation without any external localization feedback? 📸We present #LoGoPlanner, which handles perception, localization, and planning in one go! Check our results on L Addresses critical aspects of navigation and localization for mobile robots.

  Source
• AOShei/go2-quadruped-sim

  2025-12-17 - Research Publication - A complete ROS 2 Jazzy simulation package for the Unitree Go2 quadruped robot.

  Source
• jj7258/unitree_go2w_ros2

  2025-12-14 - Research Publication - ROS2 Humble workspace for the Unitree Go2W (Wheeled) robot. Includes complete 16-motor control drivers, URDF description, and kinematic simulation tools.

  Source
• tosemfdk/isaacsim5.0_ros2_go2

  2025-12-10 - Research Publication - The goal of this repository is to achieve full control of the Unitree Go2 quadruped robot within the Isaac Sim 5.0 simulation environment, utilizing ROS 2 as the communication and control framework.

  Source
• cmjang/InternNav-deploy

  2025-12-09 - Research Publication - Edge deployment guide for InternNav-based perception and navigation on Unitree Go2 / Go2W / B2 robots (ROS 2, RealSense, Python).

  Source
• Beyond Topology: A Morphological Symmetry Graph Representation for Locomotion Policy Learning

  2025-11-30 - Research Publication - Reinforcement learning has enabled impressive locomotion skills on articulated robots, but common policy representations remain only weakly aligned with robot physics. Generic networks ignore kinematic structure, while graph-based policies encode connectivity without specifying how physical quantities transform across symmetric body parts. We introduce a morphological symmetry graph representation for locomotion poli

  Source
• X-IONet: Cross-Platform Inertial Odometry Network for Pedestrian and Legged Robot

  2025-11-11 - Research Publication - Learning-based inertial odometry has achieved remarkable progress in pedestrian navigation. However, extending these methods to quadruped robots remains challenging due to their distinct and highly dynamic motion patterns. Models that perform well on pedestrian data often experience severe degradation when deployed on legged platforms. To tackle this challenge, we introduce X-IONet, a cross-platform inertial odometry

  Source
• Kaweees/unitree_mjlab

  2025-10-30 - Research Publication - MjLab external project for Unitree Go2 robot

  Source
• FishPlusDragon/unitree-go2-slam-toolbox

  2025-10-21 - Research Publication - slam-toolbox, visual robot model, PointCloud2_to_Laserscan, ROS2

  Source
• Risk-aware reinforcement-learning paper evaluates adaptive quadruped locomotion on Unitree Go2

  2025-10-16 - Research Publication - In this work, we study risk-aware reinforcement learning for quadrupedal locomotion. Our approach trains a family of risk-conditioned policies using a Conditional Value-at-Risk (CVaR) constrained policy optimization technique that provides improved stability and sample efficiency.

  Source
• Bridge the Gap: Enhancing Quadruped Locomotion with Vertical Ground Perturbations

  2025-10-15 - Research Publication - Adds a real-platform Go2 locomotion artifact for robustness under dynamic terrain perturbations.

  Source
• mehedihasan1909030/Quadruped-Robot-State-Estimation-Go2-Kinematics-Velocity-Tracking

  2025-10-13 - Research Publication - This project implements real-time state estimation algorithms for the Unitree Go2 quadruped robot using forward kinematics and sensor fusion. The system demonstrates how legged robots estimate their position, orientation, and velocity using only onboard sensors - a critical capability for autonomous navigation.

  Source
• Asset-Centric Metric-Semantic Maps of Indoor Environments

  2025-10-12 - Research Publication - Large Language Models (LLMs) can help robots reason about abstract task specifications. This requires augmenting classical representations of the environment used by robots, such as point-clouds and meshes, with natural language-based priors. There are a number of approaches to do so in the existing literature. While some navigation frameworks leverage scene-level semantics at the expense of object-level detail, othe

  Source
• grasp-lyrl/unitree_go2w_agent_sdk

  2025-10-08 - Research Publication - A unified, AI agent-friendly SDK for the Unitree Go2W quadruped robot that enables seamless integration of perception, planning, and control through a single API interface.

  Source
• Quadruped-dyn-insp/Go2_planner_suite

  2025-09-29 - Research Publication - This repository contains a complete autonomous navigation stack for the Unitree Go2 quadruped robot.

  Source
• shivam-sood00/unitree-sim2real

  2025-09-13 - Research Publication - Sim-to-sim and sim-to-real pipeline for hardware deployment on Unitree Go2 robot

  Source
• Ragtime-LAB/OpenGo2Air

  2025-09-12 - Research Publication - OpenGo2Air - An open-source hardware & software quadruped robot based on Unitree Go2

  Source
• Kodo-Robotics/go2-autonomous-patrol

  2025-09-11 - Research Publication - Autonomous security and monitoring system for the Unitree Go2 quadruped robot using ROS 2, Nav2, and a web-based control dashboard.

  Source
• wso2-incubator/unitree-go2-realtime-agent

  2025-09-05 - Research Publication - An AI agent built for the Unitree Go2 EDU quadruped robot, integrating OpenAI Realtime API with WSO2 technologies, and the Unitree Go2 SDK.

  Source
• Non-conflicting Energy Minimization in Reinforcement Learning based Robot Control

  2025-09-01 - Research Publication - Efficient robot control often requires balancing task performance with energy expenditure. A common approach in reinforcement learning (RL) is to penalize energy use directly as part of the reward function. This requires carefully tuning weight terms to avoid undesirable trade-offs where energy minimization harms task success. In this work, we propose a hyperparameter-free gradient optimization method to minimize ene

  Source
• First Order Model-Based RL through Decoupled Backpropagation

  2025-08-29 - Research Publication - There is growing interest in reinforcement learning (RL) methods that leverage the simulator's derivatives to improve learning efficiency. While early gradient-based approaches have demonstrated superior performance compared to derivative-free methods, accessing simulator gradients is often impractical due to their implementation cost or unavailability. Model-based RL (MBRL) can approximate these gradients via learne

  Source
• Bipedalism paper deploys risk-adaptive loco-manipulation policy on Unitree Go2

  2025-07-27 - Research Publication - Loco-manipulation of quadrupedal robots has broadened robotic applications, but using legs as manipulators often compromises locomotion, while mounting arms complicates the system.

  Source
• leo01110111/Quadrupeds_Climbing

  2025-05-28 - Research Publication - This is an Isaac Lab task that trains a robotic quadruped (Unitree Go2) to climb steep terrain.

  Source
• GenPO: Generative Diffusion Models Meet On-Policy Reinforcement Learning

  2025-05-24 - Research Publication - Recent advances in reinforcement learning (RL) have demonstrated the powerful exploration capabilities and multimodality of generative diffusion-based policies.

  Source
• Glowing-Torch/Deploy-an-RL-policy-on-the-Unitree-Go2-robot

  2025-05-19 - Research Publication - This repository provides a framework for low-level control of a legged robot (Unitree Go2), using ROS 2 as the communication middleware. The MuJoCo simulator is used to validate the control policy in simulation. Once the policy performs well in MuJoCo, you can deploy it on the real robot by simply setting the ROS parameter is_simulation to false.

  Source
• illusoryTwin/go2-ros2-control-sim

  2025-05-11 - Research Publication - ros2 control implementation for Unitree Go2 quadruped robot with classic control and RL-based control

  Source
• lpigeon/unitree-go2-mcp-server

  2025-05-09 - Research Publication - The Unitree Go2 MCP Server is a server built on the MCP that enables users to control the Unitree Go2 robot using natural language commands interpreted by a LLM.

  Source
• pengzhenghao/robot-docker

  2025-03-14 - Research Publication - The tutorial and docker image to play with Unitree Go2 quadruped robot dog.

  Source
• h-naderi/unitree-go2-slam-nav2

  2025-02-20 - Research Publication - Integration of SLAM and autonomous navigation on the Unitree-Go2 robot.

  Source
• wxyanxw/Go2-EDU-Mycobot-320-M5-Gazebo-Simulation-with-Navigation-and-MoveIt

  2025-01-14 - Research Publication - This repository provides a comprehensive Gazebo simulation environment for the Unitree Go2 EDU quadruped robot and the Mycobot 320 M5 robotic arm. The simulation integrates ROS Navigation Stack for Go2 EDU's autonomous navigation and MoveIt! for Mycobot 320 M5's motion planning and control.

  Source
• alexeiplatzer/unitree-go2-mjx-rl

  2024-11-17 - Research Publication - Reinforcement Learning quadruped locomotion for a Unitree Go2 robot in Mujoco XLA (MJX)

  Source
• inria-paris-robotics-lab/go2_odometry

  2024-11-13 - Research Publication - Simple state estimation for the unitree Go2 robot.

  Source
• lez666/Unitree-Go2-GestureDetection

  2024-11-11 - Research Publication - Based on Unitree Go2 python sdk , mediapipe library to achieve gesture reconization and the robot dog will react accordingly.

  Source
• svarshneysjsu/Go2Bot-OpenAI-Integration

  2024-08-24 - Research Publication - This repository showcases the integration of the Unitree Go2 robot with OpenAI, developed during a summer research project. It features voice command processing, and AI-driven task execution. Built with Python, Flask, and JavaScript, this project enhances human-robot interaction and explores advanced robotics applications.

  Source
• anujjain-dev/go2_description

  2024-08-19 - Research Publication - This package contains the robot description model for the Unitree Go2 robot for configuration with the CHAMP controller in ROS 2 (humble). It includes development of robot description model for ROS 2 Humble distribution build over unitree_ros package.

  Source
• anujjain-dev/unitree-go2-ros2

  2024-08-18 - Research Publication - Developing robot description model for Unitree Go2 robot configured with Champ Legged Robots Research Repository

  Source
• fan-ziqi/robot_lab

  2024-07-06 - Research Publication - The source registers exact model-specific Isaac Lab environments for tracked quadruped, wheeled-quadruped and humanoid platforms.

  Source
• tfoldi/go2-webrtc

  2024-02-13 - Research Publication - WebRTC API for Unitree GO2 Robots

  Source
• zeonsunlightyu/LocomotionWithNP3O

  2024-01-30 - Research Publication - unitree Go2 robot learns locomotion with N-P3O algorithm and HIM alike policy trained by isaacgym

  Source
• machines-in-motion/Go2Py

  2023-10-27 - Research Publication - A Python interface and simulation environemtn for the Unitree Go2 robot.

  Source
• elijah-waichong-chan/go2-convex-mpc

  Research Publication - Optimizes ground-contact forces with centroidal model-predictive control, Raibert-style foot placement and a separate swing-leg controller. Runs the exact Go2 model in MuJoCo; the reported forward, lateral and turning motions are simulation results. Releases MIT-licensed controller code, Go2 MJCF and URDF models, runnable trot examples and performance plots.

  Source
• RCILab/RCI_quadruped_robot_navigation

  Research Publication - Integrates reinforcement learning, navigation, and Open-RMF for autonomous driving with Unitree Go2 and Go2W.

  Source
• CURT1S03/quadruped-drl-platform

  Research Publication - A platform for training a Unitree Go2 quadruped robot via PPO in NVIDIA Isaac Lab.

  Source
• gcairone/exploration_go2

  Research Publication - Combines LiDAR mapping, Nav2 navigation, frontier selection and a custom 3D-to-2D occupancy-grid projector in a ROS 2 pipeline. The physical setup specifies Go2 EDU, an NVIDIA Jetson and Livox Mid-360 LiDAR; its Gazebo experiments instead use a TurtleBot3 model. Provides exploration nodes, map-projection code, configuration files and separate simulation and real-robot launch instructions.

  Source
• khaledgabr77/unitree_go2_ros2

  Research Publication - ROS 2 Jazzy integration for the Unitree Go2 quadrupedal robot using the CHAMP controller framework.

  Source
• Unitree Go2 sim-to-sim validation: Isaac Lab to MuJoCo

  Research Publication - Shiawase Mono Kobo reports transferring an Isaac Lab-trained Unitree Go2 walking policy to MuJoCo through ONNX. The report explains joint-order mapping, observation construction and PD-control settings, and publishes its inference-loop script. After correcting the mapping, it reports stable straight walking in a flat-ground simulation with a 0.5 m/s command. Tests of turning, rough terrain and physical hardware remain future work.

  Source
• Full Autonomy Stack for Unitree Go2

  Research Publication - Combines Point-LIO-based mapping, terrain traversability assessment, collision avoidance and FAR route planning for Go2 EDU. Uses the robot’s built-in L1 LiDAR and IMU, with waypoint, assisted-joystick and manual modes; documents both Unity simulation and physical-robot operation. Releases the autonomy stack and launch scripts, while documenting LiDAR noise, low-obstacle detection limits and occasional SLAM drift.

  Source

Media library

• Go2 app control

  image · Unitree Robotics

  Source
• Go2 on rock terrain

  image · Unitree Robotics

  Source
• Go2 sensor and hardware layout

  image · Unitree Robotics

  Source

Official product specifications, developer resources and exact-product imagery for the Unitree Go2 quadruped robot.

• Go2 product and technical reference

  WEB · Go2 · Unitree Robotics

  Source
• Unitree SDK2

  WEB · Go2 development · Unitree Robotics

  Source
• Unitree ROS 2 support

  WEB · Go2 ROS 2 development · Unitree Robotics

  Source

• Akari and Oriental Shiraishi validate autonomous quadruped walking on suspended scaffolding

  2026-09-08 - Demonstration - LiDAR detected hanging chains and site equipment for obstacle avoidance and route planning. The controller used body tilt and ground-reaction information to maintain posture across uneven surfaces. The robot detected sheets or cables snagging its legs and autonomously moved to free them; routine site patrol remains in development.

  Source
• Unitree launches Go2 embodied-AI quadruped robot

  2023-07-12 - Product Launch - Unitree unveiled Go2 as a new-generation consumer-grade quadruped robot with 4D LiDAR and embodied-AI positioning. The release placed Go2 below industrial quadrupeds in price while adding AI and perception upgrades over earlier consumer robot dogs.

  Source

Market context

Compact quadruped robot with 4D LiDAR and 360° perception.

• Target environment: Research labs, university robotics programs and light-duty inspection environments.
• Workflow context: Inspection, research, development and light operational tasks using onboard perception and autonomous navigation.
• Customer context: Robotics researchers, developers and inspection teams seeking an accessible quadruped platform.
• Deployment model: Commercial product sold as a development and light-operations platform.
• Commercial maturity: Commercially available quadruped with broad developer adoption due to accessible pricing.
• Adoption constraints: Compact form factor limits payload and terrain capability compared to industrial-grade quadrupeds.
• Market position: Unitree's accessible quadruped platform, widely used for research and light operational use. Sits below the heavier-duty B2.
• Adjacent products: B2, Spot, ANYmal

Comparable Systems

• Lite3

  Deep Robotics - Quadruped

• MagicDog

  MagicLab - Quadruped

• A2

  Unitree - Quadruped

• ANYmal

  ANYbotics - Quadruped

• Unitree As2

  Unitree - Quadruped

Related Systems

• LYNX M20

  Deep Robotics - Quadruped

• LYNX M20S

  Deep Robotics - Quadruped

• LYNX S10

  Deep Robotics - Quadruped

• MagicDog-W

  MagicLab - Quadruped