ABB RobotStudio and Training Robots in Simulation

From wind tunnel models to ABB RobotStudio HyperReality: how simulation and synthetic data train factory robots, and why home robots will learn the same way.

INDUSTRIAL ROBOTICS

Chat With Robot

9/23/20261 min read

Engineers used to test almost everything physically. NASA put scale models of the Space Shuttle in wind tunnels to see how they behaved. For robots, testing meant building a real cell, running it and fixing whatever crashed, one expensive attempt at a time.

Small model of the Space Shuttle mounted inside a round wind tunnel test section
Small model of the Space Shuttle mounted inside a round wind tunnel test section

A Space Shuttle orbiter model in a NASA wind tunnel, 1980. Photo: NASA / Wikimedia Commons (public domain)

ABB's RobotStudio moved much of that work onto a laptop more than twenty years ago. It runs the same control software as the real robot, so engineers can lay out and program a cell and check it for collisions before anything is bolted to the floor. ABB says about 60,000 customers use it.

Engineer typing on a laptop beside a white ABB industrial robot arm
Engineer typing on a laptop beside a white ABB industrial robot arm

An engineer programs a robot cell in RobotStudio next to an ABB industrial robot. Photo: ABB

AI-driven robots now need huge amounts of training data. On September 1, Oslo startup Kinetic Blocks opened a beta marketplace where companies buy and sell humanoid training data, including teleoperation recordings and first-person video. Real examples are slow to collect, which is why simulated data matters more every year.

Small grey humanoid robot with a glowing face panel standing on a carpet
Small grey humanoid robot with a glowing face panel standing on a carpet

A Unitree G1 humanoid, the kind of robot trained on large datasets. Photo: Sayanesy / Wikimedia Commons (CC0)

ABB's tool for this is RobotStudio HyperReality, built with NVIDIA Omniverse and announced in March. It renders photorealistic scenes with changing light, reflections and misaligned parts, and produces synthetic images to train robot vision. ABB says Foxconn reached 99 percent accuracy on assembly tasks this way before its robots touched the real line.

Two engineers with a laptop and tablet showing the same robot cell design near an ABB robot
Two engineers with a laptop and tablet showing the same robot cell design near an ABB robot

Engineers share a robot cell design through RobotStudio Cloud. Photo: ABB

RobotStudio Cloud lets a team edit the same cell design at the same time from different places, and AR and VR views let engineers walk around a virtual cell before it exists. ABB plans to release HyperReality to all RobotStudio users in the second half of this year.

Engineer wearing an AR headset gesturing next to an ABB GoFa cobot
Engineer wearing an AR headset gesturing next to an ABB GoFa cobot

Using an AR headset to check a GoFa cell before it is built. Photo: ABB

Home robots will need the same approach, since no company can let a robot practice in thousands of real living rooms. My bet is on a digital copy of your home, built from a phone scan, where a robot rehearses tidying up before it tries the real thing.

Living room with a grey sofa, yellow armchair and brick wall
Living room with a grey sofa, yellow armchair and brick wall

A home a future robot might first learn to tidy in simulation. Photo: Justin Schüler / Unsplash via Wikimedia Commons (CC0)

Innovation

AI solutions for effortless ABB robot control.

Automation

Robotics

ceojohntran@chatwithrobot.net

+84905311611

© 2025. All rights reserved.

qtran1215@gmail.com