ROBOGEN SKENDE
Platform

Walker S2

A life-size industrial humanoid. Its stature and reach match an adult human — which is why it works in a plant designed for people.

176cm
Height
69kg
Weight
52
Degrees of freedom
15kg
Payload
7.2km/h
Walking speed
225cm
Arm span
Specifications

Full specification

Weight~70 kg
Joint servomotors42
Payload, both arms15 kg
Grip, one hand7.5 kg
Walking speed7.2 km/h
With load≤ 2.5 km/h
Navigation accuracy±2 cm
Compute (vision)275 TOPS
Batteries2× 48 V / 7 Ah LiFePO4
Runtime2–3 h
Battery swapautonomous, ~3 min
Consumption (full day)< 10 kWh
Operating temperature0–40 °C
Humidity20–80%
Transportflight case, 170 kg with robot
ControlRC remote · web UI · API
Mechanics · motion

34 joints, each with its own motor

These figures are not illustrative — they are read directly from the robot's kinematic model in the developer kit. Motion is driven by 42 integrated joint servomotors.

34 joints, each with its own motor
Our Walker S2 · photo ROBOGEN
Degrees of freedom
Arm (excl. hand)7 × 2
Leg6 × 2
Waist · head2 · 2
Dexterous hand11 × 2
Total52
Torques
Waist pitch265 Nm
Hip and knee225 Nm
Shoulder80 Nm
Elbow45 Nm
Wrist · ankle20 · 65 Nm
What this means
The payload limit is set by the arms, not the legs. Hips and knees deliver 225 Nm per joint at a body weight of 69 kg — under vertical load the robot has several times the reserve.
Reach

Arms longer than human

A 225 cm span at 176 cm height. The robot reaches over a quarter further and deep into crates or across a pallet.
Deep squat

Lifts from the floor

The waist pitches 90° and the knees fold into a full squat. The working envelope reaches from the ground up to 1.8 m.
Sensing

How the robot perceives its surroundings

From its cameras and sensors the robot builds a real-time 3D picture of the scene and feels both force and touch in its grip.

6 cameras2× stereo (depth), 2× fisheye (wide-angle), waist RGBD front, back RGBD rear
Inertial measurement units (IMU)pose and balance in real time
Six-axis force sensorsin both arms — force and torque in the grip
Tactile sensors6 arrays per hand — the robot feels touch
Visual navigation (VSLAM)space mapping to ±2 cm
Detail of the robot's dexterous hand
Dexterous hand of the platform · photo ROBOGEN
Hands and grip

A dexterous hand and custom grippers

The robot has dexterous five-finger hands for both a delicate grip and a firm hold. Where an ordinary hand is not enough, purpose-built grippers are fitted — and this is the area developed by MH Designs and manufactured by Tech Production.

  • 11 degrees of freedom per hand
  • 6 tactile sensor arrays — the robot feels contact
  • 7.5 kg grip with one hand, up to 1 kg with the fingers
  • Custom grippers tailored to a specific part
Perception · control

Two brains: motion and vision

The robot has two control computers. One handles motion and balance in real time, the other perception of the surroundings and decision-making. This is exactly the layer that STU's development plugs into.

Motion · motion control
  • Intel Core i7 (x86), Linux with a real-time kernel
  • Real-time control of 34 joints
  • Inertial measurement units (IMU) for balance
  • Six-axis force sensors in the arms
Vision · perception and decision-making
  • NVIDIA Jetson Orin — 275 TOPS, ROSA 2 system
  • Stereoscopic vision — depth in real time
  • 2× RGBD depth camera, wide-angle cameras
  • VSLAM visual navigation with ±2 cm accuracy
Depth map from the robot's cameras
Depth map · source: UBTech
Sees distance

The colour map is a real output from the robot's cameras — every point in the scene has a distance assigned. From this the robot works out where the edge of a part is, how it is oriented and where to reach.

Navigation accuracy±2 cm
Compute performance275 TOPS
Developer kitROS 2 · Isaac Sim
Programming · integration

An open platform for developers

The robot has two programming layers — a high-level API to launch tasks and low-level ROS 2 for full motion control. This is exactly the layer STU develops.

  • ControlCenter API (JSON-RPC) — launch skills, navigation and speech over the network
  • ROS 2 Humble — control of 34 joints (500 Hz), hands, cameras and speech
  • Developer mode — full control over the motors for custom algorithms
  • URDF model + Isaac Sim / RViz — simulation and task rehearsal before deployment
  • TTS / ASR speech — the robot speaks and understands the spoken word
  • ROSA 2.0 containers — tasks driven by a behaviour tree
Walker S2 at the battery swap cabinet
Swap cabinet · source: UBTech
Operation · power

Swaps its own battery in 3 minutes

Two 48 V / 7 Ah batteries with a 2–3 hour runtime. A robot running non-stop all day consumes under 10 kWh — energy is not a cost item that decides anything.

3 min

A battery swap with no operator intervention — the difference between a machine you have to watch and one that runs across shifts on its own.

Deployment

What the robot needs in operation

Walker S2 works in a plant built for people — just a few practical conditions to meet, which we assess right at the workstation analysis.

Safe distance1.5 m from people while moving
Workspace~4 m² (2×2 m) per robot, aisles ≥ 1.5 m
Floorflat (≤ 5 mm/m²), load ≥ 150 kg/m², dry
Environment0–40 °C, lighting ≥ 200 lux
Networkgigabit, latency ≤ 100 ms
Operatortrained — training is included
Let's talk

Do you have a job for the robot?

We are happy to show you the robot in operation and design a deployment for your production line — from workstation analysis to a working application.