We’ve been exploring ways to bring the Intrinsic platform closer together with ROS — the Robot Operating System, which acts as a backbone of all kinds of robotics products and projects around the world. The open source community has long been at the edge of pushing for greater interoperability in robotics, with ROS playing a critical role in many of these advances. Today at ROSCon 2024 in Odense, Denmark, we’re highlighting how ROS capabilities can be integrated with Flowstate, our development environment, to unlock sophisticated, adaptive solutions for industry.

‍

In collaboration with our innovation partner Comau, a leader in delivering advanced industrial automation products and systems, for the first time we’ve made it possible to integrate existing ROS-based services with Flowstate without requiring modification to the original code. This is achieved by developing a custom Flowstate skill, or software component, designed to directly interface with ROS 2 nodes developed by Comau. 

‍

Our teams have made it possible to use Comau’s MI.RA/Depalletizer skill within the Intrinsic developer ecosystem, in addition to leveraging skills that are included with Flowstate like collision avoidance or ensuring compatibility with different camera and robot brands. This modular approach simplifies the adaptation process and offers a flexible, scalable solution for diverse automation needs. These benefits have been validated by the technical experts at Comau, who understand the in-depth challenges of industrial environments.

‍

At a high level, this means we’re making meaningful progress in enhancing our platform’s interoperability with commonly used development tools, empowering developers to create intelligent solutions needed by manufacturers.  Being able to combine ROS capabilities with Intrinsic capabilities like this in one developer environment gives solution builders new ways to develop intelligent solutions.

‍

Flexible robotic depalletizing with AI-enabled vision systems 

‍

Billions of boxes are loaded and unloaded every year throughout the lifecycle of a product, from production to delivery. Getting the boxes off pallets is known as "depalletizing" and is traditionally done by a mix doing heavy lifting with limited help from robots. Traditionally, robotics solutions here have been inflexible and hard programmed to only pick certain size boxes, from a certain position, in a certain way. So while robotics has been used in logistics for this for many years, it’s still expensive, inflexible and difficult to manage in practice. 

‍

Automated depalletizing, the process of using robots to unload boxes from pallets, has typically struggled to handle diverse shapes, sizes, and positions. Traditional robotic systems often require customized setups for specific pallet configurations, making it challenging to adapt to new or mixed packaging types without incurring costly reprogramming or hardware adjustments. In partnership with Comau, and by leveraging ROS, we’ve developed an adaptive solution that can help make depalletization more flexible and efficient.

The ability to determine the 3D positions of boxes on pallets relies on MI.RA — Machine Inspection Recognition Archetype, a family of flexible vision systems developed by Comau to enable quality inspection and robotic guidance tasks in smart and seamless ways. Comau used ROS to build out their MI.RA algorithms since the modular design of ROS 2 Nodes provides an easy way to develop and debug specific tasks. 

‍

Then, utilizing both Flowstate and the ROS 2 SDKs, we created a skill that acquires sensor data from depth cameras and queries MI.RA with this data over ROS, to obtain 3D poses of boxes on the pallet. Combined with other skills in our Flowstate catalog, we are able to populate our simulated environment with box locations, and run automated grasping and motion planning to unload the boxes. The ability to seamlessly combine ROS and Intrinsic skills this way enables the reuse and repurposing of valuable code, saves development time, and helps keep the developer’s focus on creatively solving the tasks. Parameters for querying ROS can be adjusted conveniently from the Flowstate frontend, making it easy to tune MI.RA for any setup.

‍

Additionally, Flowstate users now benefit from the rich tooling in the ROS ecosystem, including RViz, Rosbag, and ROS logs. These tools, when used alongside Flowstate, provide enhanced visualization and debugging workflows. For example, during the development of the depalletizing solution, we utilized RViz to visualize sensor data and internal states from the MI.RA algorithm in real time, ensuring precise box detection and picking.

‍

With this latest collaboration, Comau has become our first industry partner to help prove out not only an Intrinsic-to-ROS communication but also a workflow for future solutions. We’re excited to build more and better ROS integrations that can bring new opportunities to solution builders in industry.

‍

Happy ROSCon 2024!

‍

As ROS developers gather in Odense, Denmark this week for the annual ROSCon conference, we also wanted to congratulate and thank the open source robotics community for their contributions to the most recent ROS 2, Gazebo, and Open-RMF releases. In addition to these updates, including many technical improvements for seasoned users, they also make it even easier to get started and to integrate new devices, concepts, and algorithms in simulation. The Intrinsic team is here at ROSCon, sharing more about this milestone and enjoying the event. Please say hello, we hope to see you!