Open platform, swappable actuation
Muscle anchor points and actuator interfaces are reserved, so the structure can migrate from mechanical drive toward muscle traction without a redesign.
Nexa · Specifications
From mechanical humanoid to bionic human body.
Nexa is the platform layer: a full-size body whose skeleton, soft tissue, face and sensing are designed against human anatomy rather than around it.
Homo SilicaFemaleHomo SilicoMale
- Full-size body
- 1:1
- Degrees of freedom
- 30
- Facial DOF
- 39
- FACS action units
- 38/60
Human skeletal proportion
Bionic-spine configuration
Mouth · eye · brow · nose · tongue · neck
Reverse-engineered from 40 facial muscles
Bionic skeleton
The skeleton is the ordering principle
It decides how the body takes shape, how it moves naturally, and how muscle and skin wrap it. Everything above it depends on getting it right.
Anthropomorphic architecture
Built on real human proportion and joint placement, so it can carry continuous muscle contours and natural motion.
Integrated, high density
Joint modules, muscle, skin, control, harness and thermal loop are integrated into the skeleton, keeping mechanical structure off the surface.
30 DOF · bionic-spine configuration
- Shoulder
- 3 × 2
- Elbow
- 1 × 2
- Forearm
- 1 × 2
- Wrist
- 1 × 2
- Cervical spine
- 3
- Thoracic spine
- 1
- Lumbar spine
- 2
- Hip
- 3 × 2
- Knee
- 1 × 2
- Ankle
- 2 × 2
Dexterous face
Reverse-engineering 40 facial muscles
Expression is not a texture problem. Each muscle is studied for its motion effect, then a lattice structure is designed backwards from that effect to form a micro actuation unit — so expressions emerge from muscle, the way they do in a face.
Lattice bionic muscle units
Facial musculature surveyed in full, then lattice structures designed in reverse from each muscle's motion effect, driving facial detail in coordination.
Soft-tissue bionic skin
Skin, soft-tissue layer and muscle keep their anatomical relationship, reproducing features such as nasolabial folds and cheek volume — the substrate for complex micro-expression.
Skin & muscle
From anatomy to lattice, in four steps
- 01
Human structure → robot mapping
Motion is decomposed into axis, amplitude, traction direction and surface change, which determines the muscle groups that participate. Each muscle module gets explicit origin and insertion points on the skeleton, placed near their real anatomical positions so load travels along the muscle.
- 02
Fully parametric muscle design
An in-house cellmap system performs lossless hexahedral fill of any muscle. Measured human elastic-modulus data per body region sets the physical target; lattice type and strut-diameter gradient are then tuned to hit it. Scan → hex fill → lattice selection → elastic modulus → TPU.
- 03
Lattice muscle, bionic shaping
Real skeletal muscle is the medical reference for direction, origin, insertion and volume distribution. That is translated into a linear lattice and anchor system that preserves directionality and force transmission, then assembled onto an in-house metal endoskeleton.
- 04
Skin scanning and automated make-up
Scan-based geometric modelling captures real human texture for the transfer-printed texture layer. An in-house automated sectional make-up system reproduces skin texture to millimetre fidelity.
First-generation lattice muscle targets form, elasticity, damping and passive bulge. Active force generation is a separate track for the hand, face and respiratory regions.
Stack
Bone, muscle, fascia, fat, skin — with tactile, pressure and temperature sensing embedded through the stack.
- 01
Bone
- 02
Muscle
- 03
Fascia
- 04
Fat
- 05
Skin
Perception system
Whole-body multimodal sensing
A bionic sensing network spanning vision, touch, force, position, motion and physiological signal — fused, aligned, and resolved into a single body-state estimate.
- 01
Multimodal coverage
Vision · touch · force · position · motion · physiology
- 02
Fusion and alignment
Spatiotemporal synchronisation, cross-modal mapping
- 03
State estimation
Pose · contact · load · balance · motion trend
- 04
Capture and reconstruction
Motion reconstruction, closed-loop feedback
Sensor layer
- Touch / force
- High-density tactile array · six-axis force sensor
- Vision
- RGB-D / stereo · first-person view
- Audition
- Microphone array · bone conduction
- Position / deformation / motion
- IMU / positional markers · deformation sensors
- Physiological
- Heart rate · respiration · EMG · surface temperature · thermal distribution
Perception map
Output · unified body-state representation
- Whole-body pose
- Contact distribution
- Load state
- Balance and stability
- Motion trend
- Interaction state
Nexa is scored, not described
Every claim on this page maps to a layer of the Biofidelity Benchmark, and every layer is measured independently.