R1-A7 Smart H — upper-body humanoid with tendon-driven dexterous hands, wrist RGB cameras & Orin Nano 40 TOPS

What's in This Configuration

The R1-A7 Smart H brings biomimetic tendon-driven actuation to desktop humanoid research. You get the full upper-body dual-arm humanoid with 7-DOF arms, waist rotation, binocular vision, voice interaction, and open SDK — now paired with tendon-driven dexterous hands that replicate the musculoskeletal architecture of the human hand. Unlike motor-in-finger designs, tendon-driven hands route actuation through flexible cables (tendons) to remote motors, enabling lighter fingers, more natural compliance, and backdrivability that mirrors human muscle behavior. Wrist-mounted RGB cameras deliver close-up visual feedback for grasp verification and hand-eye coordination. The built-in Orin Nano 40 TOPS compute module powers real-time perception and AI inference locally. At just ~13 kg, the R1-A7 Smart H is the most biomimetically authentic desktop manipulator in the R1-A7 lineup — designed for researchers who need to study and replicate human hand mechanics.

Dual 7-DOF Arms. Redundant Kinematics. Human-Like Dexterity.

The R1-A7 Smart H is engineered around two 7-DOF high-performance arms, each driven by low-inertia inner-rotor PMSM motors with crossed-roller and double-row ball bearings. With 555 mm reach and a ~2 kg payload per arm, it handles precise assembly, pick-and-place, and manipulation research with human-like fluidity. The extra two degrees of freedom per arm (compared to 5-DOF platforms) provide kinematic redundancy — the arm can maintain end-effector pose while reconfiguring its elbow to avoid obstacles, reach around corners, or optimize for natural motion. Dual encoders on every joint deliver closed-loop precision down to ±0.1 mm.

Dual 7-DOF arms with tendon-driven dexterous hands for research-grade biomimetic manipulation

Tendon-Driven Dexterity — Biomimetic Actuation for Authentic Human Hand Mechanics

The tendon-driven dexterous hands on the R1-A7 Smart H represent a fundamentally different approach to robotic manipulation. Instead of placing motors directly in the fingers, tendon-driven designs route actuation through flexible cables (tendons) from remote motors in the forearm to the finger joints — exactly like the human musculoskeletal system. This architecture delivers several unique advantages for biomimetic research. First, lighter fingers enable faster, more agile motion with lower inertia. Second, natural compliance arises from cable elasticity, allowing the hand to absorb impacts and adapt to object geometry without explicit force control. Third, backdrivability means external forces on the fingers are transmitted back through the tendons to the motors, enabling impedance control strategies that closely mimic human muscle behavior. Fourth, coupled joint motion — where pulling one tendon affects multiple joints — replicates the biomechanical coupling found in human fingers, making this the ideal platform for studying human hand mechanics, developing biomimetic controllers, and transferring insights from human motor control to robotics.

Biomimetic Grasping — Human Muscle Mechanics on Your Desktop

With tendon-driven hands, the R1-A7 Smart H performs biomimetic grasping that authentically replicates human hand behavior at the mechanical level. The tendon transmission creates inherent mechanical compliance — the fingers naturally yield to external forces, wrap around irregular object geometries, and distribute contact pressure without explicit force sensing. This passive adaptability is a hallmark of human grasping and is difficult to achieve with rigid motor-in-finger designs. The coupled joint architecture means that finger motions emerge from tendon pull patterns rather than independent joint commands, enabling researchers to study synergistic control — the same principle that allows humans to control dozens of hand muscles with just a few neural commands. The compact desktop form factor puts this biomimetic capability within reach of any research lab studying human motor control, neuromuscular biomechanics, or bio-inspired robotics.

Wrist-Mounted RGB Cameras — See From the Gripper

Integrated RGB cameras on each wrist provide a gripper's-eye view of the workspace. These close-up feeds are invaluable for visual servoing, grasp verification, and fine manipulation tasks where the head-mounted stereo pair lacks resolution or the right viewing angle. With tendon-driven hands, wrist cameras capture the unique visual signature of tendon-actuated finger motion — cable routing, joint coupling, and passive compliance — providing essential visual feedback for biomimetic control algorithms and tendon tension estimation.

Vision-Wrist Fusion — Two-Scale Perception for Biomimetic Manipulation

The R1-A7 Smart H fuses stereo vision from the head with close-up RGB feeds from the wrist cameras into a unified perception stack optimized for biomimetic manipulation. The binocular camera provides global scene understanding, object detection, and coarse grasp planning across the full workspace. The wrist cameras deliver high-resolution local views of the tendon-driven hand-object interaction, capturing finger configurations, cable routing, and passive compliance behavior in detail. Together, these two visual scales create a complete manipulation perception pipeline: the head camera finds the object and proposes a grasp strategy, the wrist camera guides the tendon-driven fingers into position, and the Orin Nano 40 TOPS processes both streams in real time to execute precise biomimetic grasps. This dual-scale visual architecture is essential for research in visual biomimetic manipulation, tendon tension estimation from vision, and imitation learning from human hand videos.

Waist Rotation — Expand Your Workspace

The 1-DOF waist joint provides ±150° of rotation, dramatically expanding the robot's reachable workspace without moving the base. In a seated or standing deployment, the waist allows the robot to turn and manipulate objects on either side, hand objects from one hand to the other across the body, and orient toward humans or workstations. Combined with 7-DOF arms and tendon-driven dexterous hands, the R1-A7 Smart H achieves human-like reach envelopes with biomimetic grasping across a full semicircle of desktop space — ideal for collaborative tasks, bimanual assembly workflows, and human-robot interaction studies.

Orin Nano 40 TOPS — AI Onboard, No Latency

The R1-A7 Smart H ships with a built-in NVIDIA Jetson Orin Nano delivering 40 TOPS of AI compute. This means visual perception, grasp pose estimation, tendon tension modeling, and reinforcement learning policies all run in real time locally. Object detection models process head camera frames, finger keypoint detectors analyze wrist camera feeds, and biomimetic control policies execute with sub-millisecond latency — all without external GPU servers or cloud dependency. Just plug in power and start training biomimetic manipulation policies.

Binocular Vision & AI Perception

A binocular camera head delivers a 146° × 124° field of view with a 60 mm stereo baseline. RGB streams at 1280×720@30 Hz and depth at 544×448@10 Hz give the R1-A7 true spatial awareness. Paired with an 8-core high-performance CPU + 10 TOPS NPU in the head module, plus the Orin Nano 40 TOPS body compute, it runs visual perception, object detection, and gesture recognition locally — no cloud required.

Voice Collaboration & Natural Interaction

Four-array beamforming microphones and dual 3 W speakers enable clear voice capture and synthesis in noisy labs or classrooms. The open audio framework supports custom wake words, multilingual TTS, and integration with LLMs for conversational command-and-control.

Compact. Lightweight. Desktop-Ready.

At just 835 × 357 × 190 mm and approximately 13 kg, the R1-A7 Smart H fits on a standard lab bench or classroom desk. The compact footprint leaves room for objects, tools, and sensors around the robot. Power via the included external supply — an optional upper-body Li-ion battery is available for ~1.5 hours of untethered operation.

Compact 13 kg upper-body design — deploy on any desk or lab bench with biomimetic tendon-driven dexterity

Full-Stack Open Development

Unitree provides a mature, Linux-based robot development framework with open APIs for the underlying system, robotic arms, audio, lighting, and visual control. ROS 2 native support, dual encoders on every arm joint, and hollow internal wiring make the R1-A7 Smart H a clean slate for reinforcement learning, teleoperation research, biomimetic control algorithms, and embodied AI. The built-in Orin Nano 40 TOPS handles model training and inference for visual perception and grasp planning without external hardware. The open SDK includes full control over tendon tensions and joint states, enabling research into impedance control, variable stiffness, synergistic grasping, and other biomimetic control strategies that leverage the unique mechanical properties of tendon-driven actuation.

Upgrade Path — Grow With Your Research

Start with Smart H and expand as your projects evolve. The R1-A7 platform is fully modular:

• End Effectors — Swap tendon-driven hands for BrainCo Revo 2 Basic or Tactile 5-finger hands, LinkerBot O6 or O6 Tactile 6-DOF hands, Dex3-1 or Dex3-1 Tactile 3-finger hands for alternative dexterity paradigms.
• Tactile — Add tactile sensor upgrades compatible with tendon-driven hands for multi-modal biomimetic manipulation research.
• Large VLM — Deploy vision-language models for natural-language manipulation commands, visual question answering, and semantic scene understanding.
• Compute Module — Swap the Orin Nano for the Orin NX 100 TOPS module to unlock 2.5× AI performance, 16 GB RAM, and dual DLA cores for simultaneous head camera, wrist camera, and biomimetic control processing.
• Upper-Body Battery — Add the Li-ion battery pack for ~1.5 hours of cord-free operation.

Designed For

Technical Specifications