November 10, 2021

MujinPalletizer QA hero400x275Japan-based Mujin recently established a U.S.-based division north of Atlanta to expand its offerings to a North American audience. The industrial robotics company is looking to expand the use of its flagship product, the Mujin Controller, an all-purpose intelligent robotic control system that uses real-time perception, motion planning and universal control to create robots that can handle complex logistical tasks.

The controller aims to manage any robotic application by guiding the movement of any robot arm via machine intelligence, automatically managing potential downtime scenarios through autonomous motion planning and perception without the need for human intervention. At the recent Pack Expo, the company demonstrated a Mujin robot and gripper that handled cases, slip sheets and pallets to build multiple pallets in a single stack, which then transferred the pallet to an autonomous mobile robot (AMR) to move to a breakdown area.

Robotics World recently chatted with Josh Cloer, director of sales for Mujin Corp., about the need for better training methods around industrial robots, the state of industrial robots within logistics, and the challenges of entering the U.S. robotics market.

Robotics-World: The goal of the Mujin Controller is to handle more complex logistics tasks for industrial robots. Can you give examples of tasks that Mujin can handle that other teaching or training methods tend to struggle with?

JoshCloer MujinCloer: With traditional robot systems, robots are guided through their programming language to specific waypoints where they interact with the workpieces for an application. Typically you will see some type of fixturing and sensing to ensure that the work pieces are in the appropriate location and orientation to allow for reliable picking by the gripper on the robot. 

In the past, what is stopping further proliferation of robotic systems is the fact that the robot arm is only as repeatable as the operations around it, and in order to take on tasks with more inherent variability the system must have some type of vision and real-time motion planning.  This is what Mujin solutions offer to end users. Robotic systems that can combine advanced sensing and real-time motion planning to handle more variability that exists within their process.

Many great examples come to mind from the logistics space, where traditionally robots have been difficult to use. If you have many different box sizes that you need to pull from a pallet or pack onto a pallet, or you have products that need to be pulled from storage within a box or tote or packed into a box or other packaging for order fulfillment, there is an inherent randomness to these types of applications that a robot cannot explicitly be programmed to fulfill. The system needs vision to understand where the workpiece is, what the orientation of the workpiece is, and determine if it is pickable. It needs sensing on the gripper to understand how it has picked an item, that it has picked a single item and how much this item weighs. It also will need to understand the constraints of the environment it operates within and make inferences for optimization and recovery given the sensory data available to it.

In short, these applications require a level of intelligence that cannot be programmed, but rather use data points and constraints to fulfill that operation in real-time, much like a human does. 

R-W: What are your thoughts on new approaches for teaching, such as gesture-based learning, or even easier, no-language-needed methods? Are these better, or does the future rely on a combination of vision, perception and artificial intelligence / machine learning?

Cloer: There are many new approaches aimed at making robot deployment seamless, but we are focused on adding capability as well as offering a ‘teachless’ or ‘no-language-needed’ approach. Our approach provides much more value and flexibility than any teach-based programming because you are not teaching the robot how to move, but rather giving the robot information about its surroundings, building relationships between areas of interest and auxiliary equipment, and allowing it to navigate within the constraints that it is given. 

With this model-based approach, you get the most optimal movements with a running log of the state of the entire work cell, which makes troubleshooting a breeze. Also, if something changes about your workflow, it is as simple as moving an object within the model to the new location or sharing the new CAD file of a workpiece.

R-W: When you talk with customers in the field, where are they on their automation journey? Have they implemented robotics and are looking to expand, or are they still at the early stage of automating manual tasks? For those customers with robots in the field, what are their pain points?

Cloer: Robots have long been adopted in process-oriented markets like automotive manufacturing, which really is the early adopter of industrial robotics. Outside of manufacturing, the adoption of robotics has been slow until the last few years. In large part this is due to the variability of their operations, which is non-conducive to traditional robotics. 

For instance, you may be surprised to hear that many of the large retailers and e-commerce companies to this day have still not deployed robot arms in their logistics operations. Intelligent robotic solutions, like Mujin’s offering, solve many of the pain points of traditional robotics, like spending long hours developing custom programs for specific projects and spending even more time to change those programs as the requirement for the site changes. 

However, the biggest pain point our solutions can solve is the growing difficulty of trying to hire workers for manual and strenuous tasks. These jobs are difficult to fill but also difficult to do, and when your job is measured by the weight you move, you crave something that allows you to use more human intuition and less human strength and endurance. Offering solutions that can fill more of these tasks helps more logistics companies build a reliable and flexible supply chain. 

MujinController400x275R-W: What was your reaction to the Pack Expo demo where you showed robot palletizing onto an AMR? Do you see these types of tasks as where the industry is going, or needs to go?

Cloer: The application was well received at Pack Expo. It is rare to find an end-to-end solution within a demonstration at a trade show event, and we wanted to show our potential customers what we think is the future of end-of-line packaging. Separating your end-of-line from your warehouse with automation boosts efficiency and safety within your operation, and can be installed within your existing work area. This allows your forklift operators to do more within the warehouse and reduce the interactions that used to be required between your workers and heavy machinery.

R-W: With the Mujin Controller, will companies be able to have robots that can perform different tasks, depending on the process they need to complete? For example, could a robot that loads/unloads pallets then be converted to a different task, such as becoming a welding/painting robot? We’re wondering whether the idea of real-time motion planning and perception could lead to a world where companies use a single robot for a number of different tasks, or whether there will still be hardware limitations.

Cloer: I believe the main limitation here is the end-of-arm tooling. The capability is there within the system to do many tasks under the same system, but each cell would need to be modeled with constraints and process flow for individual tasks. 

R-W: Mujin is known in Japan, but not as much in North America. You just expanded to open up your first office in the U.S. Does being the “new guy” present any new challenges or opportunities that the company didn’t face before?

Cloer: This is a great opportunity to bring production-ready technology into the North American market in a very flexible way. We have this startup vibe in our U.S. culture, but we also have a tremendous amount of an existing knowledge-base and resources to pull from. In some startups, marketing and messaging begin at minimum with a viable product, but at Mujin we already have industry proven technology to introduce to this market.

Learn more about Mujin by visiting its website here.

Related video: