Night shifts are hard to staff
Essential repetitive work limits operating hours because people are difficult to recruit.
Physical AI Workforce
Automate light factory workwith AI robots
Designed and assembled in-house by ManmaruAI, our AI robot automates the packing, boxing, inspection, and labeling work that factories still rely on.
01 / Overview
For light work that never justified dedicated automation
ManmaruAI designs and assembles the robot in-house. We target processes where frequent product and procedure changes made fixed automation uneconomical. A VLA model learns from demonstrations, while a low-center-of-gravity wheeled base and dual arms are designed for deployment across multiple processes.
Does this sound like your line?
Products change frequently
Dedicated equipment and repeated teaching do not pay back across changing products and boxes.
Full-time employees remain tied to repetitive work
Employees are assigned to inspection, sorting, and packing instead of higher-productivity work such as process improvement and quality enhancement.
02 / Applications
Target light-work tasks
From post-molding inspection and packing to other lightweight, indoor, repetitive work, we assess fit around actual site conditions.
Inspection and sorting
Check workpieces by camera and sort them based on the result.
Boxing and packing
Automate the sequence from grasping and alignment through placement.
Labeling
Apply labels to products or packaging with position checks.
In-process picking
Recognize, grasp, and transfer lightweight workpieces between processes.
03 / Approach
Designed for real factory use
Task learning with VLA
Learn from human demonstrations and adapt to site and product variation.
Mobile, low-center-of-gravity base
A wheeled platform designed to move between lines instead of staying fixed.
Safety first
Area-sensor detection, automatic stops, and signal lights indicate operating status.
- Payload per arm
- 1.5kg
- Designed for light packing, inspection, and labeling work.
- Repeatability
- ±0.2mm
- Stable positioning for inspection and packing.
- Arm configuration
- Dual
- One unit for grasping, inspection, and packing.
Robot demonstration
Interactive 3D model
Drag to rotate the robot04 / Process
Deployment process
We validate one process at a time, from real samples to on-site operation.
- 01
Process review
Review the workpiece, takt time, movement, safety conditions, and pass criteria.
- 02
Sample validation
Test recognition, grasping, and judgment with real workpieces.
- 03
PoC design
Design the robot, fixtures, cameras, learning data, and recovery flow.
- 04
On-site deployment
Validate operation on site and improve it with accumulated data.
05 / FAQ
Frequently asked questions
What kinds of work are a fit?
Indoor, repetitive work with lightweight objects is the main target. We review payload, takt time, floor, and safety conditions.
Does the existing line need major modification?
We start by using existing equipment. Fixtures and safety measures are defined around site conditions during the PoC.
What should we prepare for an initial consultation?
A process video, workpiece photos or samples, takt time, and work instructions help us assess fit quickly.
Physical AI Workforce