Unitree's quadruped robots (especially the Go2 series) are being purchased and deployed by various U.S. police departments as tools to enhance officer safety. These remote-controlled or semi-autonomous "robot dogs" serve as alternatives to sending humans into high-risk situations. Common uses include:

• Entering buildings to scout or "clear" areas ahead of officers (e.g., during barricaded suspect incidents, searches, or SWAT operations).
• Acting as de-escalation tools (e.g., one-way audio communication for negotiators).
• Reducing officer exposure to threats in dangerous environments, functioning like a ground-based drone with cameras, sensors, and mobility.

Specific examples from U.S. departments include:

• Pullman Police Department (Washington state): Introduced a Unitree Go2 named "MIKE" (Mission-Integrated Kinetic Entity) in late 2024 for special operations, building searches, and barricaded subjects.
• Port St. Lucie Police Department (Florida): Acquired a Unitree Go2 in 2024, deploying it in high-risk calls (including narcotics operations) to keep officers safe and clear residences before entry.
• Topeka Police Department (Kansas): Acquired a Go2 Pro model in 2025 for around $5,360.

Reports from 2025 indicate that local police departments (along with universities and even some military interest) are among the growing U.S. customers for Unitree's robots, which are significantly cheaper than comparable Western-made models (e.g., from Boston Dynamics). This has occurred despite broader U.S. political concerns about Chinese-made tech.

In China, Unitree's humanoid robots (like the G1) have been piloted in law enforcement patrol programs in some cities, integrating them into public safety and patrol duties.

Overall, the connection is commercial and operational: Unitree supplies cost-effective robotic platforms that several police agencies (mainly in the U.S. for quadrupeds) have adopted to modernize tactics and improve safety, rather than any deeper institutional partnership. Some variants are even marketed specifically for "industry & law enforcement" use.

Unitree Robotics is actively pursuing advancements in autonomy for its quadruped robots (commonly referred to as "robot dogs") and other platforms like humanoids, with a strong emphasis on AI-driven capabilities such as reinforcement learning, environmental perception, and task execution. These efforts are evident in their product roadmaps, open-source initiatives, and partnerships, which aim to enable robots to operate with minimal human intervention in complex scenarios. For instance, the Go2 quadruped features an "Advanced AI Mode" powered by large-scale simulation training, allowing autonomous behaviors like obstacle climbing, adaptive roll-overs, and 3D LiDAR-based mapping for independent navigation in predefined areas. It also includes the ISS2.0 Intelligent Side-Follow System, which uses wireless positioning for enhanced obstacle avoidance and traversal of rough terrain without constant oversight. Unitree has integrated 4D ultra-wide LiDAR and large language model (LLM) empowerment, positioning the Go2 as an "embodied AI" platform capable of decision-making and world interaction.

On the humanoid side, models like the G1 demonstrate imitation and reinforcement learning for flexible movements, including force-controlled dexterous hands for object manipulation and AI-upgraded perception. Unitree forecasts a major breakthrough in humanoid autonomy by 2026, focusing on improved comprehension, task execution, and commercial viability. The G1 already supports multi-axis compliance, inverse kinematics for motion planning, and real-time balance correction, making it suitable for AI-powered mobility in research and applications. Similarly, the B2-W quadruped incorporates hybrid locomotion (legs and wheels) for autonomous efficiency in inspections and endurance tasks. These developments build on Unitree's quadruped expertise, with algorithms transferable to humanoids, and the company collaborates with entities like Booster Robotics and Fourier Intelligence to advance AI in humanoid research.

Regarding data integration from external sources similar to Palantir (a data analytics platform focused on fusing disparate datasets for insights), Unitree's robots are designed for extensibility and tying in external data through APIs, sensors, and AI frameworks, though no direct collaboration with Palantir exists—likely due to geopolitical tensions between Chinese hardware and U.S. defense-oriented software. Instead, Unitree emphasizes open-source tools like unitree_rl_gym for reinforcement learning on robots like Go2 and G1, which supports data collection, model training, and deployment. They've released manipulation datasets (e.g., G1 hand and gripper data) and trained imitation learning models for public use in embodied AI. Partnerships, such as with Orbbec for stereo vision integration, enhance data acquisition by improving object recognition and grasping via pre-configured sensor data.

Practical examples include compatibility with ROS (Robot Operating System) for environmental data interpretation and autonomous movement. In demos, Unitree robots fetch live data from secure APIs (e.g., via OAuth2 for schedules or product info), process JSON, and incorporate it into actions like inspections. Teleoperation platforms collect high-quality motion data from human operators, serving as a "data engine" for training AI models without relying solely on simulations. Protocols like UDP/DDS and SLAM enable real-time data fusion for navigation and machine learning integration. This mirrors Palantir-like functionality by allowing robots to pull and analyze external data for tasks such as patrols or de-escalation, as seen in team deployments of Go2 dogs with humanoids. Unitree's open collaboration strategy fosters a developer ecosystem to expand these integrations, dominating 70% of the global quadruped market while prioritizing affordability and AI extensibility.

Sources on Unitree Robots and Police Departments

1. The Wire China. While Politicians Fret, Unitree is Selling Robots Across America. November 23, 2025. https://www.thewirechina.com/2025/11/23/while-politicians-fret-unitree-is-selling-robots-across-america-unitree-police-robot

2. Pullman Police Department. Meet “MIKE”: Pullman PD Introduces Robot Dog. October 30, 2024. https://www.pullman-wa.gov/dept_news_detail_T37_R228.php

3. Yahoo News. Topeka Police Department acquires new robot dog. August 7, 2025. https://www.yahoo.com/news/articles/topeka-police-department-acquires-robot-224420933.html

4. Port St. Lucie Police Department. PSLPD unveils robotic dog to assist with law enforcement efforts. Facebook post. Accessed February 2026. https://www.facebook.com/PortStLuciePolice/posts/pslpd-unveils-robotic-dog-to-assist-with-law-enforcement-efforts-acquired-throug/788078686773858

5. KREM.com. Pullman police introduce robot dog. October 31, 2024. https://www.krem.com/article/news/local/pullman-robot-police-dog/293-6831ddf2-a93f-4f83-8120-ef2632519132

Sources on Autonomy Advancements, AI Modes, and Related Developments in Unitree Robots (Go2, G1, etc.)

6. Unitree Robotics. Robot Dog Go2. Official product page. Accessed February 2026. https://www.unitree.com/go2

7. South China Morning Post. Chinese start-up Unitree sees humanoid robots in wide commercial use this decade. January 31, 2025. https://www.scmp.com/tech/tech-trends/article/3296895/chinese-start-unitree-sees-humanoid-robots-wide-commercial-use-decade

8. IEEE Spectrum. Humanoid Hype Dominates Top Robotics Stories of 2025. December 24, 2025. https://spectrum.ieee.org/top-robotics-stories-2025

9. Superhuman AI. Robotics Special: Unitree stuns the industry (again). August 9, 2025. https://www.superhuman.ai/p/robotics-special-unitree-stuns-the-industry-again

10. All About Industries. Not Humanoids, But Robotic Dogs Are the Most Important Testing Ground for Embodied AI. November 3, 2025. https://www.all-about-industries.com/not-humanoids-but-robotic-dogs-are-the-most-important-testing-ground-for-embodied-ai-a-7bda5cb754fead1431825dc992e5e9af

Unitree focuses on open-source reinforcement learning (e.g., unitree_rl_gym), ROS compatibility, sensor fusion (LiDAR, cameras), API extensibility, and collaborations like those with NVIDIA/Google for AI training or Orbbec for vision—drawn from official product docs and industry reports above.