Moz1

About

About Moz1

Moz1 is a 26-degree-of-freedom wheeled humanoid with two seven-axis arms and an omnidirectional mobile base. Force-torque and position control allow the arms to regulate contact during insertion, folding and other manipulation tasks, while whole-body collision detection and safe-torque-off functions govern operation around people.

The development stack includes ROS 2 APIs, teleoperation, autonomous lidar navigation and simulation resources for NVIDIA Isaac and MuJoCo. Developers can access whole-body control at up to 120 hertz, record robot data and adapt external models such as Pi0.5. Spirit AI supplies documentation and development tooling around the physical platform.

Capabilities

Source-backed capabilities

  • Workflows: Dual-arm manipulation, mobile navigation, embodied AI, structured task work and supervised physical assistance.
  • Operating environments: Factories, offices, service sites and controlled indoor spaces where wheeled mobility can support task execution.
  • Target markets: Industrial partners, service operators, investors and developers evaluating Spirit AI's wheeled humanoid direction.
  • Deployment model: Prototype mobile manipulator in Spirit AI's embodied robotics product line.
  • Flexible-material handling

    Pilot - 2026-02-25 - Spirit AI reports Moz1 agents handling flexible wire harnesses on CATL battery production lines with a reported success rate above 99%.

    Source

Deployments

  • Active Customer Operations

    Moz1 agents operating on CATL battery production lines - 2026-02-25 - CATL - battery production-line automation - Spirit AI reported Moz1 agents operating on CATL battery production lines.

    Source
  • Named Commercial Site

    1 named commercial deployment - 2025-12-18 - CATL - battery manufacturing - SCMP reported CATL deployed Spirit AI's Moz1 humanoid robots at the Zhongzhou battery pack line, describing a factory-floor deployment context.

    Source

Supply

Supply / Components

  • Moz1

    Mid 360 LIDAR - sensing_perception - Livox - Livox MID-360 lidar provides SLAM sensing on Spirit AI Moz1 - source: docs.spirit-ai.com

    Source
  • Moz1

    D405 Wrist Depth Camera - sensing_perception - RealSense - Two RealSense D405 cameras provide wrist perception on Spirit AI Moz1 - source: docs.spirit-ai.com

    Source
  • Moz1

    D455 Head Depth Camera - sensing_perception - RealSense - RealSense D455 serves as the head perception camera on Spirit AI Moz1 - source: docs.spirit-ai.com

    Source

Research

  • Spirit AI releases Spirit-v1.5 fine-tuning code

    2026-04-01 - Research Publication - Spirit AI updated the Spirit-v1.5 repository in April 2026 with fine-tuning code, separating the training release from the January initial model and benchmark release. The repository documents training requirements, dataset loading, distributed training utilities, and a fine-tuning launcher for adapting Spirit-v1.5 to robotic manipulation datasets. The release matters because it makes Spirit-v1.5 more reproducible as a developer resource for embodied-AI research rather than only a checkpoint and benchmark artifact.

    Source
  • Spirit AI open-sources Spirit-v1.5 robotic foundation model

    2026-01-11 - Research Publication - Spirit AI released Spirit-v1.5 as an official VLA robotic foundation model implementation with model architecture, inference wrapper, RoboChallenge integration and model checkpoints. The repository says Spirit-v1.5 ranked first on the RoboChallenge Table30 benchmark as of January 11, 2026, and links the Hugging Face model release. The release matters because it gives Spirit AI a first-party model artifact behind its embodied-intelligence stack and provides a public benchmark path for the Moz1 product story.

    Source
  • Moz1 arm paper presents analytical inverse kinematics for Spirit AI's 7-DOF robot arm

    2025-11-28 - Research Publication - The Moz1 inverse-kinematics paper presents a closed-form analytical solution for the seven-DOF Moz1 robot arm with wrist offsets. The author reports full self-motion support, singularity handling, and all 16 solutions per pose for the arm configuration.

    Source

Timeline

  • Spirit AI and JD Group form strategic partnership for retail embodied intelligence

    2026-03-19 - Partnership - Spirit AI and JD Group signed a strategic partnership agreement covering consumer product customization, technical integration, real-world deployment and joint marketing from 2026 through 2029. The first tangible outcome is Spirit AI's Moz robot going online at JD MALL physical stores for high-precision coffee brewing and service demonstrations. The deployment uses JD's JoyAI large model and JoyInside technology with remote operation, multimodal data capture, joint trajectories and force feedback to support embodied model training.

    Source
  • Spirit AI reports Moz1 agents operating on CATL battery production lines

    2026-02-25 - Deployment - Spirit AI said its VLA models have been applied to CATL production lines. The company said Spirit AI-powered agents handle flexible wire harnesses, achieving a reported 99%+ success rate while matching skilled human workers on precision and cycle time. The deployment is framed as industrial validation of Spirit AI's embodied model approach, not a disclosed broad commercial rollout.

    Source

Context

Market context

Wheeled humanoid robot with dual arms designed for real-world task execution in structured environments such as factories and offices.

  • Target environment: Factories, offices, service sites and controlled indoor spaces where wheeled mobility can support task execution.
  • Workflow context: Dual-arm manipulation, mobile navigation, embodied AI, structured task work and supervised physical assistance.
  • Customer context: Industrial partners, service operators, investors and developers evaluating Spirit AI's wheeled humanoid direction.
  • Deployment model: Prototype mobile manipulator in Spirit AI's embodied robotics product line.
  • Commercial maturity: Prototype-stage product with public product page and limited deployment proof.
  • Adoption constraints: Adoption depends on manipulation reliability, software maturity, safety, payloads, task definition and site integration.
  • Market position: Wheeled humanoid lane trading legged mobility for stability and near-term indoor usability.
  • Adjacent products: DexForce W1, Kinisi 01, Maker H01, Quanta X2

Comparable Systems