Quick Menu

A company that understands the changes in the world and challenges the future through technological innovation.

BE THE BEST, BE THE FIRST

Human Friendly Actuator (HFA) Philosophy
Human Friendly Actuator Official Logo

Why HFA?
Hardware Wisdom & Embodied Intelligence

HIGEN HFA goes beyond motor manufacturing. We design joint-level compliance and intrinsic safety so humans and robots can physically interact in shared spaces with complete peace of mind.

✓ Intrinsic Safety (Yielding) ✓ Sensorless Proprioception ✓ < 100ms Local Edge Reflex
HFA Hardware Wisdom
HFA Badge
Level 1: High-Level Control (AI Interface & Global Planning) Think & Plan
Level 2: Mid-Level Control (Efficient Communication Protocol) 20 ~ 50 Hz Sync
Level 3: Low-Level Control (Actuator-Embedded Edge AI & Physical Control) < 100ms Reflex
3 Core Philosophy Pillars HFA Logo

3 Hardware Wisdom Pillars for Collaborative Robotics

01 Reflex Arc Analogy
Body Intelligence (Hardware Reflex Arc)

Embodied Intelligence

Immediate response without waiting for central AI

Just as a human hand recoils from heat without waiting for brain decisions, HFA joints sense shock and immediately yield (<100ms).

Philosophy Detail HFA
02 Proprioception Model
Smart Force Control (Proprioception)

Smart Force Control

Compliance via sensorless torque estimation

Without expensive external torque sensors, dual 24-bit encoders estimate friction and external forces to deliver smooth compliance.

Philosophy Detail HFA
03 Human Safety First
Intrinsic Safety & Active Yielding

Intrinsic Safety & Yielding

Flexible yielding mechanism alongside humans

In accidental collisions with humans, the joint instantly lowers stiffness to absorb impact energy, protecting both human and robot.

Philosophy Detail HFA
01. The Challenge

01. Robots Need More Than Just a Smart Brain

Even if central AI processes giant VLM models, physical disturbances occur in microseconds. The joint hardware itself must yield dynamically.

Simulation-Driven Intelligence

Intelligence trained in simulation is optimized for ideal conditions.

Real-World Uncertainty

Reality is full of sensor noise, latency, friction, and environmental shocks.

Physical Response Matters

Robot safety and natural motion begin from immediate body reflexes.

Robots Need More Than Just a Smart Brain
02. Sim-to-Real Gap

The Simulation Was Perfect.
Reality Was Not.

Motions that were perfect in virtual simulation collapse under real-world friction and payload fluctuations. HFA actively compensates for dynamic errors inside the joint.

01

Parameter Mismatch

Discrepancy between simulated models & physical reality

Simulation (Ideal)

In simulation, mass, inertia, and torque remain unrealistically constant.

Physical Reality Obstacle

Physical parameter shifts due to component wear, assembly tolerance, and heat.

02

Sensor Noise & Delay

Sensor errors, latency & incomplete data

Simulation (Ideal)

Simulated data enjoys zero latency and zero measurement noise.

Physical Reality Obstacle

Real-world communication latency and noise destabilize feedback loops.

03

Actuator Nonlinearity

Distortion from friction, heat & vibration

Simulation (Ideal)

Motor output torque and gear transmission maintain ideal linearity.

Physical Reality Obstacle

Nonlinear physical realities such as friction, cogging torque, and backlash distort motion.

04

Environmental Complexity

Unpredictable variables like impacts & humans

Simulation (Ideal)

Idealized environments with flat ground and predefined objects.

Physical Reality Obstacle

Unpredictable contacts, collisions, and terrain requiring <100ms response.

03. The Wiser Body

The Answer Was Already
in the Human Body

Just as a human hand recoils from heat without waiting for brain decisions, HFA joints sense shock and immediately yield (<100ms).

Human Brain
↕ Mapping
High-Level Central AI (LLM / VLA)
Reflex Arc
↕ Mapping
Local Actuator Response (<100ms)
Proprioception
↕ Mapping
Torque, Position & Motion Sensing
The Answer Was Already in the Human Body
04. Distributed Intelligence

HIGEN Distributes Intelligence Across Three Layers

Dual encoders compare the positional displacement between the motor shaft and output shaft, while a current sensor measures changes in applied current. This data is fused through holding-torque compensation and a disturbance observer to estimate the direction and magnitude of external force without a dedicated torque sensor.

Level 1 - Think & Plan

High-Level Control & Global Planning

Full-body motion and path planning, static gravity compensation

Central Control
Level 2 - Coordinate

Efficient Communication Protocol

Bandwidth reduced to 20–50Hz, delivering trajectory goals

20 ~ 50 Hz Sync
Level 3 (HIGEN CORE) - Sense & React

Actuator-Embedded Edge AI & Physical Control (<100ms)

Torque transparency, real-time shock detection, and impedance yielding at contact.

< 100ms
Local Edge Reflex
05. Changing Role & Pillars

From Motion Component to Sensory Organ

A 3-layer architecture that slashes central communication overhead and handles physical disturbances locally at the joint level.

Conventional Actuator HIGEN Smart HFA (HFA Innovation)
Passive position/velocity command execution (Command-based drive) Real-time active detection of force, contact & impact (Real-time force detection)
Simple repetition of fixed, predefined motion (Repetitive task) Instantly adjusts torque and posture and yields
Focused on output power and precision alone Focused on compliance and intrinsic safety
Waits for a central controller's decision Instant on-site response within 100ms at the joint edge level (Level 3)

Embodied Intelligence

More than a mere power component — realizes bodily intelligence that interacts with the physical world and adapts in real time.

Smart Force Control

Perceives external force through proprioceptive algorithms, without a torque sensor, to deliver optimal, smooth force control.

Safety (Yielding)

Delivers intrinsic physical safety that yields on its own to even the light force of a child's hand upon collision.

HIGEN Human Friendly Actuator

Powering the Next Generation of Humanoid Robots