Vention has opened a Physical AI research lab in Montreal, Canada, focused on translating advances in robotic perception and manipulation into systems capable of reliable operation in manufacturing environments. Led by Dr. Jimmy Li, Director of Physical AI, the lab brings together robotics control, motion planning, classical computer vision, vision foundation models, learning from demonstration and reinforcement learning, with work directed toward industrial goods, electronics and automotive production. The program is structured around the requirements that ultimately determine whether an AI based robotic application can move beyond a research demonstration: reliability, repeatability, deployment cost and performance under changing production conditions.

The lab draws directly on Vention’s manufacturing installed base, giving researchers access to production environments, industrial manipulation data and ongoing deployment feedback. The company says its technology is used across thousands of manufacturers, including 90 of the Fortune 500, creating a practical foundation for model development and post-training. Factory applications introduce variables that controlled demonstrations cannot fully reproduce, including mixed part geometry, irregular presentation, clutter, occlusion, changing lighting and differences in process conditions. Those conditions can be carried back into perception, pose estimation, grasp selection and path planning, then tested again in live production environments.

Vention factory floor graphic illustrating Physical AI enabled robotics and industrial automation, Industrial Automation Monthly.

Vention’s new Montreal Physical AI lab is advancing robotic manipulation for manufacturing by combining industrial production data with perception, foundation models, grasp planning, motion control and factory deployment.

“Physical AI will fundamentally expand what manufacturers can automate, but the challenge is no longer simply proving that a robot can perform a task in a lab,” said Etienne Lacroix, founder and CEO of Vention. “The real opportunity is making these capabilities reliable, economical, and deployable across thousands of factories. By combining Canada’s world-class AI ecosystem with Vention’s deep robotics expertise, industrial data, and full-stack automation platform, we have a unique foundation to close that gap.”

A central part of that work is GRIIP™, Vention’s modular Physical AI pipeline for robotic manipulation. Introduced in February 2026, GRIIP spans scene digitalization, object segmentation, pose estimation, grasp selection and collision-free motion planning, combining foundation models from technology providers including NVIDIA with Vention’s own models. The company also plans to release a public GRIIP Software Development Kit that will allow engineering teams to extend the pipeline for individual applications. Vention is already applying the technology with industrial and electronics manufacturers, including work with a large automotive OEM on complex, unstructured robotic tasks in final assembly.

GRIIP treats perception and motion as one connected manipulation process. The system first builds a representation of the workspace, identifies objects and determines their position and orientation, then evaluates potential grasp points and generates a collision-free trajectory for the selected grasp. In production, each stage affects the next. Changes in part position, geometry, surface properties, lighting or surrounding objects can alter the perception result, which in turn changes grasp selection and robot motion. Vention’s research therefore spans the complete perception-to-motion sequence rather than developing vision, planning and control as isolated functions.

That software stack is also tied directly to Vention’s factory floor control architecture. MachineMotion AI combines industrial motion and safety functions with NVIDIA Jetson processing for AI based robotics and machine vision, alongside EtherCAT motion control, networking and I/O. The controller can manage robots, motors, conveyors, sensors and other automation hardware while providing local accelerated computing for Physical AI applications. Connecting model execution with the same control environment used to run the machine gives Vention a direct route from perception and planning into physical robot motion rather than requiring a separate AI computing layer to be integrated afterward.

The Montreal lab is intended to keep that development cycle closely connected to manufacturing. Production challenges can shape research priorities, while new models and manipulation capabilities can be returned to customer environments for evaluation under operating conditions. Dr. Li has led Vention’s Physical AI strategy and technical collaboration with NVIDIA over the past two years, with work spanning robot perception, autonomous manipulation and model driven motion planning. Dr. Joelle Pineau, Chief AI Officer at Cohere, has joined the program as External Technical Advisor, contributing to model architecture decisions, research prioritization and connections with the broader AI research community.

“What makes this lab different is the loop we’ve built: academic research feeding directly into live production problems, and production feedback feeding back into the research,” said Dr. Jimmy Li, Director of Physical AI. “Our clients aren’t waiting for a finished product to test; they’re in the room while we build it. That’s unique in this field, and it’s what lets us move faster from a research result to something that actually runs on a factory floor.”

The work targets manufacturing applications where conventional automation becomes difficult to maintain as variability increases. Fixed robotic sequences remain highly effective when part location, orientation and process conditions are tightly controlled. Applications such as deep bin picking, machine tending, kitting, depalletizing and selected assembly operations introduce a different requirement because the robot has to interpret the current scene before it can determine how to act. Vention has been applying GRIIP across these types of workflows, using a common perception-to-motion architecture rather than building a separate vision and planning stack for every individual task.

Vention says Physical AI is now its fastest-growing business segment and reports that related revenue increased 400% during the past year. The company has not disclosed the underlying revenue base, so the percentage remains a company-reported growth measure rather than an indication of the segment’s absolute size. The lab currently has eight team members, with Vention reporting more than 16 additional openings across robotics control, simulation and Physical AI research as the program expands.

For industrial automation, the technical threshold is repeatability. A manipulation system has to maintain acceptable cycle time, accuracy and recovery behavior across changing parts and production conditions, not simply complete a difficult task in a controlled demonstration. Vention is building its Montreal lab around that requirement, combining industrial data, model development, perception, planning and factory control within the same development loop as it moves Physical AI toward broader manufacturing deployment.

About Vention

Founded in Montreal, Canada, in 2016 by Etienne Lacroix and Max Windisch, Vention develops an integrated industrial automation platform spanning machine design, simulation, programming, motion control, robotics hardware and deployment software. Its platform combines cloud-based engineering tools, modular automation hardware and edge control so manufacturers can design, program, deploy and operate robotic cells and other automated equipment within a common environment.

Vention’s technology portfolio includes MachineBuilder™ for machine design, MachineLogic™ for automation programming and simulation, MachineMotion AI™ for industrial motion and edge computing, and a growing portfolio of robotics and Physical AI applications. MachineMotion AI incorporates NVIDIA Jetson processing with EtherCAT motion control, machine vision connectivity, networking, safety and I/O functions, providing a common hardware layer for conventional automation and AI driven robotic applications.

The company reports more than 28,000 Vention machines deployed worldwide across thousands of factories and more than 25 manufacturing industries. Its applications span robotic assembly, machine tending, material handling, palletizing, sanding, packaging and other factory automation processes. Vention is headquartered in Montreal and opened its European headquarters in Berlin in 2021, supporting manufacturers across North America and Europe as the platform expands from modular automation into software defined and Physical AI based robotics. For more information, please click here. 

Source/Photo Credit: Vention

(Editor’s Note: All trademarks mentioned in this article, including company names, product names, and logos, are the property of their respective owners. Use of these trademarks is for informational purposes only and does not imply any endorsement.)

Molly Bakewell Chamberlin
Latest posts by Molly Bakewell Chamberlin (see all)
Tagged