Will the pandemic accelerate automation in supply chains? Part 2: Cheaper, smaller, more efficient – the evolving capabilities of robots
Robotics was once the domain of big manufacturers capable of installing multi-million dollar robots in production lines but that’s all changing as it becomes easier to build, install and operate robotics in supply chains
It’s a quiet revolution but it’s there. Pull up the shutters of distribution facilities and open the doors of factories and you can find it going on, progressing more rapidly than ever before. I am of course talking about the new wave of robotics entering the supply chain.
For most people, when their delivery drops on their doorstep, there is little thought of how that happened and all too often it’s just a grumpy ‘about time’ that crosses people’s lips. Peel back the curtain though and most would be staggered by the systems that now sit in the warehouses of Amazon and Ocado, with robots sweeping across the floor in a tireless ballet.
We are only just emerging into a new age of automation across supply chains right from the top of the production process down to the final delivery to the end customer. Robotics as a field is going through a period of exponential growth that will transform not just our jobs or shopping habits, but our entire world.
In part two of our feature on automation (read part one here), we asked the experts at the frontline of this revolution how and why this happening and what the consequences will be.
A new breed of robots
When we watch television programmes from the not too distant past that tried to predict our future, robots were supposed to be ubiquitous and highly advanced by 2020, with humanoid automatons walking amongst us.
The problem is that it is extremely challenging to bring together all of the complex subsystems necessary to recognise objects, understand an environment, make even the most basic of choices and undertake complex manipulation. We as humans tend to neglect the complexity of even some of our more mundane tasks.
The very simple task of grabbing an item, scanning a barcode, somehow confirming that that's the right product that was supposed to be picked and then putting it into one either box one, two or three, depending upon what's been ordered, and know for sure that you did it. That's really state of the art. That's hard
For example, RightHand Robotics is trying to replicate the capacity of the human hand to pick out and handle all of the objects we find in modern picking processes. Company Head of Product and Marketing, Vince Martinelli admits to those looking in from the outside, “The task that robot was designed for is absolutely the simplest thing in the world,” however it needs “all kinds of advanced vision and gripping and math to make things work and algorithms, machine learning - the whole laundry list of cool technology,” to make it work. “It also has some cognitive skills that it has to master. You want to start putting these all together to do the very simple task of grabbing an item, scanning a barcode, somehow confirming that that's the right product that was supposed to be picked and then putting it into one either box one, two or three, depending upon what's been ordered, and know for sure that you did it. That's really state of the art. That's hard.”
Critically, the challenge of replicating the capabilities of the human hand can now be grasped and handled (pun intended) due to the coalescing of a number of different factors that are making robots more effective from both a functional point of view and a cost perspective.
Advances in material sciences and lower cost manufacturing techniques, such as 3D printing, have reduced the costs of many critical parts, from the vision systems, to the internal processors, to the batteries, to the actuators - all have been falling precipitously in price.
The advancements in physical systems are meeting a steady stream of advancements in Artificial Intelligence (AI) and machine learning. These are even more important when it comes to the wider spread and use of robotics in supply chains. They are making robots more efficient and adaptable, widening their use cases and allowing for more operations to be undertaken in a safe, efficient manner. As noted in part one of our piece by Lior Elazary, Founder and CEO of inVia Robotics, they can reduce the need for exacting mechanical precision through software and hardware combining to sense accurately the environment.
Elazary also notes how “Software allows us to operate in different workflows. And what that means is that if you say somebody needs a particular item … and that workflow is very, very unique to this warehouse,” the software “allows us to adapt to the existing environment,” to the extent that their customers can use templates “that we use to guide them, as to what's the best practices, but then they just tune it to their, warehouse: ‘I have a post here. I have a conveyor belt here.’ Then the robots learn that passage.”
Robots can be easily reconfigured very rapidly. Most of our customers are able to reconfigure their entire workflow with the robots in under an hour
This adaptability is helping “to open the doors to more automation to companies,” believes Elazary “because what's happening is that [end users] are really paying for the productivity of the robots, as opposed to the robots themselves [as they were] in the past.”
Melonee, CEO of Fetch Robotics, agrees that this more modular and configurable approach is a game changer for further deployments of robotics in logistics. Their “Robots can be easily reconfigured very rapidly. Most of our customers are able to reconfigure their entire workflow with the robots in under an hour,” which has proved extremely important at this particular juncture. “Most of our customers went from one to two shifts a day, or to three to four shifts a day to enable social distancing, because they couldn't have as many people in the building,” which was a gap robots were best placed to fill and that they can now fulfil more easily as they can be deployed more rapidly, into more situations and at far lower cost than they could have been 10 or 15 years ago.
They can start to enjoy double or triple productivity within four to six weeks
For Karen Leavitt, CMO for Locus Robotic, “The most important thing for an operator to understand is this: That it really is a very easy, low threshold for starting the process of adopting robotics technology. They don't have to think of this as a massive undertaking. In fact, it can be done relatively simply, and they can start to enjoy double or triple productivity within four to six weeks,” claims Leavitt.
This doesn’t mean it’s all plain sailing, however, as new challenges and complexities continue to arise. “Because of COVID, we've been asked to go into increasingly more difficult environments with more traffic, or spaces,” notes Wise. “It's kind of like … autonomous cars. There's a definitely a huge difference between traffic in a sleepy town and a big city and the highway, and I think in facilities, you see the same things. We're getting pushed more and more to go to more and more difficult environments.” Wise sees “a lot of our technology challenges are around” this area of operating in more dynamic environments. “How do we make sure that in all of these very diverse environments, with different rules of the road and different behaviours and different expectations, that we provide a product that lives up to all of those different expectations?”
Leavitt notes that although “Many factors are consistent across the board for all of our customers,” they also need to “take into consideration the characteristics of the specific operation. What types of goods are they handling? Are they picking? Are they large and bulky? Is their low physical volume or is there a high volume of merchandise? Do those volumes change during the day? Understanding all of those characteristics allows us to put together what we call a concept of operation for each site, where we understand what the workforce has to be doing at any given hour of the day, any week of the year, any month, any season,” underlining that software is the critical underpinning to robotics rapidly expanding role in supply chains.
The need to consider each situation and build the right tools is especially important for those looking at robotics towards the upper funnel of production in supply chains, where environments can be a lot rougher and more unpredictable than a distribution centre.
We don’t try to shoehorn a robot into being able to do a job. If there is a demand there, then FANUC will make a robot that is specific for that demand
For FANUC UK, they are looking at putting robots to work in the field. “We’ve got a robot now in trial up in Scotland picking broccoli off the back of a tractor,” says Tom Bouchier, Managing Director, for FANUC UK. “So, we have to make it waterproof, so that it has the right shielding, that the vision camera can move fast enough to process at the speed a tractor is moving over the top of a field for it to be able to pick,” amongst other challenges.
Indeed, the capacity to take adaptable underlying systems and then apply the latest manufacturing techniques means that companies can now make robots almost to spec for the client’s situation. “I think what is changing now is that we don’t try to shoehorn a robot into being able to do a job. If there is a demand there, then FANUC will make a robot that is specific for that demand,” says Bouchier.
Installing from afar
COVID-19 is also playing a prominent role in accelerating the tendrils of automated colleagues into our workspaces. As noted in part 1 of our piece, there are significant push factors in the need to meet e-commerce demand and increase productivity to cope with higher costs and social distancing measures. Unexpectedly though, these social distancing measures are in fact making it easier to put in place a robotics operation.
Automation companies have been forced to think about how to make it as easy as possible to set up an operation and to allow the end-users to do much of it on their own with low training burdens. The long-term implication of this will be wider usage as another barrier to adoption becomes easier to clear.
When “March 2020, came around we already had commitments to do deployments into customer buildings,” says Fergal Glynn, VP of Marketing at 6 River Systems. “So, we accelerated our investment into the tools and technologies and then also local support to help with the deployment. Fast forward three months later. Now we are doing deployments today in multiple countries, such as the US and in Spain, where robots are getting shipped to a 3PL’s warehouse. There it's an employee of the 3PL who un-crates the robot. It's the employee of the 3PL who will push around the robot to create the map. It's the employee of the 3PL who is able to do whatever physical work is required in the building and then we've got our deployment engineers and software engineers, because of all the sensors and cameras on the robot, able to gather all the information they need in order to get our robot working in that building.”
If you look at the things that COVID has brought to the door, it's definitely usability
Martinelli likewise notes that they’ve “Been able to do a couple things. One is remotely support projects where we might have normally had some people on the ground, so the customer can keep running the system and production. We've handed over more of the responsibility to them, or in some cases to partners we're working with, and have improved training and documentation to make it such that other people could step in and help us when we can't physically get to a site.”
I think in the past, if we tried to convince our customers to deploy remotely, they would have got scared
Wise points out that “If you look at the things that COVID has brought to the door, it's definitely usability, because we are deploying our technology more and more remotely. We've had to invest more there,” so that “someone who has barely any knowledge of computers can learn how to train a robot.
This has meant a sea-change says Elazary. “I think in the past, if we tried to convince our customers to deploy remotely, they would have got scared. They would have asked, ‘Well, what do you mean you're not coming on? Why can't you help with this?’ Now, it's a different question. Now it is ‘what can you do without going there?’”
What does this mean for workers?
Although there is a major expansion of automated capacity in supply chains, it doesn’t mean that we as humans are out of the picture yet. Far from it in fact.
Humans need to be kept in the loop for a number of different tasks and will play a critical role for some time to come.
At a basic level on the production line, “Fine manipulation is still a challenge, especially if you try to pack,” says Elazary, which is often “a very challenging piece for robotic technology.” This is especially so when it comes to non-uniform goods that need a degree of quality control.
If you have a robot that cost a million dollars, it's very difficult to justify against a person who costs a lot less and can do that very quickly, very efficiently
For example, Markus Schmidt, President of Swisslog Americas notes that “A vision system will see an apple and see it is round and knows how to grip it, but if it is kind of rotten or a little bit out of shape, it will not easily [comprehend] this” compared to a human.
“If you have a robot that cost a million dollars, it's very difficult to justify against a person who costs a lot less and can do that very quickly, very efficiently,” explains Elazary.
I think we, as an industry, are looking at automating, probably around 75% of the workload, and then the other 25% will remain conventional for a while
Although companies like RightHand Robotics are making significant strides towards this capability and will eventually supplement or replace humans in these roles, it will be some time yet.
“I think we, as an industry, are looking at automating, probably around 75% of the workload, and then the other 25% will remain conventional for a while,” believes Schmidt.
However, this is just the most basic level, and in fact it is going to be critical to revamp the whole workforce, with a significant focus on upskilling.
“Robots have got a reputation for replacing workers,” says Bouchier but “I think what [supply chain operators) are finding now is that the upskill of that labour is much more important, as is their permanent labour force. A lot of these companies utilise temporary labour and other ways to get through the peak demands and they’re realising that automation can do that but only if they upskill their workforce.”
How do we use our current labour to fulfil many, many more orders?
This means taking talent from within the facility and using their knowledge to supervise their robotic colleagues, thus making robots a force multiplier.
Customers are asking inVia, “how do we use our current labour to fulfil many, many more orders?” says Elazary. “What we do in particular, is we convert workers into what we call ‘robot wranglers’,” which are upskilled workers still at the frontline of the distribution centre but they “are not necessarily pickers. They're still fulfilling orders, but they're managing the robots,” in conjunction with inVia’s “robotic operation centre, where people are monitoring the robots.”
Robots are not able to solve their own problems. They really need people to go in and solve their problems, but they're really good at doing that same task, over and over and over and over again
Similarly, RightHand Robotics has operators utilising control centres, where what were once warehouse workers now “Have a console with screens, through which they're getting alerts and information from a fleet of robots that might be scattered over the facility. So, there's always a human element to the to the system…. Now the same person is the supervisor of the robot team.”
“This is really where people shine,” says Elazary. “Robots are not able to solve their own problems. They really need people to go in and solve their problems, but they're really good at doing that same task, over and over and over and over again.”
There is a huge change management component if you're retooling an existing business
Looking further up the chain though, “There is a huge change management component if you're retooling an existing business,” warns Terrie O’Hanlon, CMO at GreyOrange. “Because software is not subject to the human foibles of perception,” and potentially allows teams to see “very objectively what to do,” it is managers who might be finding themselves with the most fundamentally changed roles in supply chains.
O’Hanlon gives an example from their software “where it might be that Jim is actually better at picking and packing larger items … than if he's doing cosmetics or chapstick or these smaller items…. Statistically the system will show that and then help you assign the right people to the right picking tasks overall, which a normal manager like couldn't really discern,” as quickly or in as much detail. Automation will therefore come with adjustments throughout the workforce and middle and middle-upper management will also need retraining and reskilling to work with systems, software and statistics skills.
Fundamentally however, this is all about productivity Bouchier reminds us. “We’re doing a big system that we just put into North America. It’s making the same product with the same number of product but it is making four times the output. Nobody is getting replaced, it is just that they are able to be more competitive in the marketplace. It’s a matter of increasing your output and improving your quality.”
Why not sign up to our newsletter and be of all of our great content?
Upcoming Virtual Roundtables & Webinars to Sign Up For:
Wednesday July 15, 2020 – 4PM BST / 11AM EST
Thursday 16th July – 4PM BST / 11AM EST