SciTransfer
HYPER · Project

AI Brain for Industrial Robots to Handle Unpredictable and Complex Tasks

manufacturingTestedTRL 5

Imagine a robot that learns to move and touch things just like a child does, by trying and getting feedback. Instead of following a rigid script, it builds its own internal map of the world to understand how to react to changes. This means it can handle messy or changing environments where humans are currently the only option.

By the numbers
80-100%
success rate in robot cube transfer task
30-65%
baseline success rate (ACT)
10-20
times lower loss compared to baseline
The business problem

What needed solving

Industrial robots are currently too rigid for unpredictable environments, while existing AI lacks the reliability and scalability needed for mass production. This forces companies to rely on expensive human labor for complex, variable tasks.

The solution

What was built

An end-to-end AI/AGI robot brain and an API for controlling KUKA, ABB, and Universal Robots hardware.

Audience

Who needs this

Warehouse automation providersHigh-mix manufacturing plantsRobot OEM integratorsE-commerce fulfillment centers
Business applications

Who can put this to work

Logistics
enterprise
Target: Warehouse Operator

If you are a warehouse operator dealing with unpredictable package shapes and sizes — this project developed an AI robot brain that enables 80-100% success rates in object transfer. It allows robots to handle variability without needing manual code patches for every new item.

Automotive Manufacturing
enterprise
Target: Assembly Plant

If you are an assembly plant dealing with rigid automation that fails when parts are slightly misaligned — this project developed a foundation model that controls joint angles and gripping force. This provides the flexibility to handle objects of different weights and fragility.

Electronics
SME
Target: Small-scale Component Assembler

If you are a component assembler dealing with high-mix, low-volume production — this project developed an API that integrates with KUKA, ABB, and Universal Robots. It allows for faster convergence and lower loss in task execution compared to standard AI models.

Frequently asked

Quick answers

What is the cost or pricing model for this AI?

Based on available project data, pricing and cost structures are not specified.

Can this be scaled to a full industrial plant?

The project aims to solve the lack of scalability found in current hybrid AI workarounds by providing a foundation model that maintains robustness while adding flexibility.

How is the IP or licensing handled?

Based on available project data, specific licensing terms are not provided, though the project is led by an SME (SICSAI AB).

How does it integrate with existing hardware?

The system includes an API that enables control of robots from major OEMs including KUKA Robotics, ABB, and Universal Robots.

What is the timeline for deployment?

The project period is from 2025-07-01 to 2026-12-31.

Consortium

Who built it

The project is led by a single Swedish SME, Superintelligence Computing Systems SICSAI AB. With a 100% industry ratio and no university or research partners, the project is heavily driven by commercial application and rapid development rather than academic exploration.

How to reach the team

Contact Superintelligence Computing Systems SICSAI AB in Sweden

Next steps

Talk to the team behind this work.

Contact us to explore integration of this AGI model into your robotics fleet.

More in Manufacturing & Industry 4.0
See all Manufacturing & Industry 4.0 projects