What’s New in AI for Robotics… From Simulation Twins to Smarter Integration

As we approach the second half of 2026, the excitement around AI and Physical AI continues. The latest developments target traditional programming, deployment times and digital simulation training – essentially using AI to make robots easier to deploy, simulate and scale.

Smarter Integration Powered by AI

This month, Robotiq announced IQ: an AI-based software platform for robotic palletising integration. This allows Robotiq’s partners to take a 3D scan of a customer’s site, using LiDAR (Light Detection and Ranging) and IMU (Inertia Measurement Unit) technologies on an iPhone Pro. Ceiling height, obstacles, conveyor positions, aisle widths and floor space are all captured in one pass, connected directly to simulation software. What used to require hours of manual measurement and follow-up visits gets done in a single pass on site – saving time, avoiding manual-entry errors and speeding up deployment.

Read more: https://blog.robotiq.com/how-iq-gives-manufacturers-a-faster-more-predictable-path-to-a-running-palletizing-workcell

Digital Twins and Simulation Are Becoming Key

Typically, a challenge with Physical AI training is that it relies on expensive real-world data collection, with slow iteration cycles and the risk of damaging hardware. Simulation offers a digital way to gather vast amounts of synthetic data and train AI models, as well as continuously improving systems after deployment. A digital twin is a high-fidelity virtual replica of a physical robot, sensor or system that behaves like the real thing in simulation. It is not just a 3D model – it combines physics, sensor behaviour and environment interaction. Another announcement that came from Robotiq this month was the release of its TSF-85 tactile sensor fingertips as a digital twin on NVIDIA Isaac Sim. The fingertips, which are compatible with proven Robotiq grippers, can detect pressure, vibration and proprioception. Essentially, this enables developers to simulate not just what a robot sees - but what it feels.

Read more: https://blog.robotiq.com/robotiq-releases-tsf-85-digital-twin-on-nvidia-isaac-sim

The Move Towards No/ Low Code Programming

The Universal Robots brand has always been known for its ease-of-use and accessibility for users with no prior robotics experience. This is being taken a step further by AI platforms – such as Acteris by Trener Robotics – which replace point-to-point programming with Vision-Language-Action (VLA) models enabling natural language task specification. The robot builds ‘skills’, rather than relying on scripted programming, which can be executed and scaled across multiple cells. As well as its simple, chat-based interface, Acteris is a repeatable software layer that can cope with high-mix environments and real-world production conditions.

As we see AI tools impact on all areas of robotic automation, from simulation training in development to site scanning for deployment, it is evident that there is a shift away from manual, rigid processes. These advancements are making robot deployment faster, improving accessibility to robot programming for users of all skill levels and driving forward Physical AI training with digital twins that mimic real-world physics and objects.

Looking ahead, we are likely to see more robotic systems that can be trained, tested and refined, rather than programmed. For manufacturers and integrators, this means better flexibility, lower complexity and the ability to scale automation more widely than before. In that sense, robotics is becoming less about programming machines and more about training systems that continuously improve.