0:03
Hello everyone and welcome to today's webinar, which is on optimising commercial strategy and productivity with automation, digital services and vision systems, which is hosted by our PA members and back.
0:15
First of all, thank you to everyone for coming today. We hope that you find the content useful.
0:19
There will be a Q &A at the end, so please put any questions that you have into the questions box and we'll do our best to go through as many as we can at the end of the webinar.
0:26
and following today's webinar we'll be sending you an article or email in the next couple of working days and this will contain a recording of the webinar, a copy of the slides and the contact details of today's presenters.
0:36
So without any further ado, I'll hand you over to today's presenter, Tom.
0:41
Hey, thank you very much Luke.
0:43
So as you can probably see we're sharing a screen at the moment with a couple of slides, I just wanted to do some brief introductions to everyone that's on the panel and we've already covered the overview.
0:55
So today you can have myself as the integration director across the impact group.
1:00
There's Michael Lewis as well, who is our digital solutions product manager and Ed Straighten who is our head of vision systems.
1:07
And as you can tell from the photos on the screen, the other two have a much greater understanding about how to take a photograph than me.
1:13
So you're in safe hands with the subjects we're gonna cover today.
1:17
To give a bit of an understanding about how we're planning on running this webinar.
1:20
We're going to run it and try to address the key areas that are stated on the webinar brief and kind of a round table format where I think my role is to try and triple bed and Michael as much as possible and forgive me I forgot to mention we were due to be joined by Christian Turian as well and however we are pleased to announce that he's currently off work well just after welcoming his first child so he won't be joining us today but I'm sure if you have any questions, we can make sure he follows up on them afterwards.
1:50
There'll be a couple of polls during today's webinar, so we're going to kick off with one.
1:54
The idea of the polls is just to gauge what level you guys are at in terms of your automation and digital journey, and also just to provide us a little bit of feedback so that we can try and tailor some of the conversation around what's particular to you guys that are dialed in.
2:09
Then as introduced by Luke, we've got a questions and answers section at the end.
2:15
So please let's try and challenge Ed and Michael as much as we can do throughout this whole thing.
2:19
That's why I'm here to watch them squirm, which will be good.
2:23
So with that being said, any further ado, we wanna start with the first audience poll, which is what stage are you currently at in integrating digital and machine vision technologies to support automation?
2:38
We've put a couple of brackets there.
2:40
We would request for you guys to start putting your answers as they go along, but we'll review the findings just after the first subject as well, so we can discuss what level everyone's at there.
2:53
Thank you very much for filling that in in the background.
2:55
I believe that should be live now so that you can cast your votes now.
3:02
First subject we've got is addressing the application of automation and digital solutions, and in particular in optimising commercial strategy.
3:12
Everyone probably is on a different point in their journey on this but the aim of it is what's the impact that automation can have and how do we leverage the digital tools to increase profit, increase sales, decrease costs, lots of different aspects that can be addressed in this.
3:27
So I'll throw it over to Ed first, What's your thoughts?
3:33
Thank you, Tom.
3:35
So yeah, I think it's really important when we talk about digital or automation, we can mean so many different things.
3:44
And we really, we find the most success when you consider the context.
3:49
So in a lot of organizations, we see very similar challenges.
3:52
You've got kind of market pressures, you've got cost control, you've got your regulatory requirements.
3:59
there's ever expanding customer expectation and there's a load of complexity that you're often dealing with.
4:07
The question's no longer about, should we automate?
4:10
It's about where do we do it?
4:12
How do we do it?
4:13
And how much do we do it?
4:16
And in all honesty, there's no universal answer to that question.
4:19
It depends entirely on your goals, the maturity of the processes you've got and how much you want to change.
4:27
So to get started on that journey, it really isn't about technology, it's about asking the right questions.
4:35
And before you invest in that digital or machine vision solution, you've got to understand kind of what challenge you're addressing, where are your bottlenecks in processes, have you got the data that backs that up?
4:49
Are you trying to reduce waste or are you trying to improve throughput, do more with less or reduce your errors?
4:55
or do you have customers in the supply chain wanting you to check things and it's mandated?
5:01
There's a lot of different reasons why you might be entering into this.
5:05
And fundamentally, you need the right data to support it.
5:08
And when you have that right data, is it being used effectively?
5:12
Is it just on a dashboard or are you actually using it?
5:15
And are you confident in that data that you're capturing?
5:19
And who's gonna use it at the end of the day?
5:23
We've seen a number of projects, they enter difficulty when you buy a solution and you look for a problem.
5:31
What I'm keen on talking about is more about, well, what's the problem before we start?
5:37
And then really getting under the skin of what that is, and then we can look to, well, how do you effectively automate to meet your commercial objectives rather than to just make a process better?
5:49
But I don't know, Michael, what have you seen kind of work effectively on this?
5:53
Yeah, I mean, just to pick up on some of the points that you raised, a lot of my, what I'm going to discuss is going to be very technology focused.
6:00
But here you are absolutely right to say we need to put technology to one side right at the start of the kind of scoping the problem statement.
6:07
And let's not run the risk of being led by an exciting new technology.
6:12
Let's first understand and define the problem.
6:14
And I think, you know, ensuring that the right people are involved at that definition stage and making sure that we're capturing the information in the right way and making sure that stakeholders are understanding of what individual people's problems are is incredibly important.
6:28
You know, simple things like capturing user stories and making sure that we're understanding the problem in a somewhat objective way that people can interpret.
6:39
In involving the right people as early as possible.
6:41
For me, as a technology-focused IT person, it is always IT, and I find that discussing people with, sorry, discussing problems with organizations, I often find that they want to include IT as little as possible and as late as possible, and that can really scupper a problem or a technology initiative.
7:00
And just to describe some of the work that we're doing, so MPAC produced a number of different white papers with all of these kinds of problem statements in them, and anticipating what different stakeholders, what information they might need at what point, and we put those together in various different white papers.
7:19
You already mentioned something there, Michael, about what an MPAC could offer.
7:22
Now, obviously, we all work for MPAC, but due to confidentiality agreements, we can't share intricate details of some of the projects we've done.
7:30
So if we talk in a bit more of a generalized tone, it's because specific applications, we do keep confidential because they are potentially differentiated for our customers.
7:40
You also mentioned there about really defining what the challenge is at the beginning.
7:45
I think if we just scroll on to the next bit, which is the poll results, which I have open on my screen ahead of me.
7:51
So if I look at that, for those of you that have voted, we've got roughly a third of the people are in the we are exploring or planning to start phase.
8:00
And then two thirds of the people are in that we have pilot projects or partial implementation.
8:06
so the majority of the audience that we're talking to now are in the early stages or they've made that choice to go for it and will have probably been through that defining what the problem statement is and trying to really get to the bottom of what it is that you're trying to achieve and it kind of flows on to the second part of apologies there is a second poll that we're going to ask again now and then we'll come on to it so and second poll is with everyone who's being in charge of an early stage project, what are the problems and the challenges that you guys are facing in particular?
8:42
And I'm sure that the options we've set out there cover 95% of the things, so if you could cast your votes now that would be really useful so that we can start to again tailor some of the conversation as it goes along.
8:56
And I'll sort of lead in while you're all voting, I'll lead on to the next next process we've got where both Michael and both mentioned there about, you know, almost put the technology to one side, define what the problem is, and then figure out how you're going to solve it with the tools that are available.
9:14
But if we close that poll now, thank you Luke, thank you, and something appears to have happened to my screen, I'm not entirely sure what everyone can see now. We can see slides again at that point.
9:45
Well, my screen's crashed, but we'll go on and you'll just have to deal with our faces.
9:49
I'm very sorry, but they were just prompts anyway.
9:51
So the next subject that we were hoping to look at is exploring the link between digital solutions, machine vision, automation, and productivity improvement.
10:01
And I kind of want to tie it back to what Ed and Michael said in terms of shelter technology at the moment, because there's so many phrases that are coming on at the moment.
10:12
which things such as Digital Twin, which I think I've spent the last three or four years trying to explain it means something different for everyone because there's multiple applications.
10:22
AI, you can't open your phone without some form of AI tool supporting you or AI supposedly the answer for everything.
10:30
I suppose we hand that question over to the man in charge of digital solutions, but there's a lot of phrasing in technology.
10:37
What are the links between them and how do you make the most out of them?
10:40
Yeah, so the first response I was going to say I wasn't going to geek out, I might geek out a little bit now and I do pick up on the IAI term and the literal links between different data sources.
10:53
What we're seeing internally, we're running a few different pilot projects for our own internal IT systems.
10:59
We're also working with our customers on how do we take data from X location, CRM, and combine that with all the data sources that we have in the business.
11:07
And historically, they would be mammoth tasks that would require perhaps hundreds of thousands of pounds projects to get consultants in to understand how do we link data source from location A and location B.
11:21
It is really exciting what's happening with the development of AI and how those tools can be used to link systems together.
11:27
And as I mentioned, we have a couple of pilot projects that we're running internally, which are incredibly interesting and exciting.
11:33
At the click of a button, we can get data from source A, combine it with source B and produce something new.
11:40
And those kinds of new technology in the horizon are really something that we should start paying attention to and something that certainly I'm excited by.
11:49
And I think Ed is certainly excited by too with the integration pass that we could potentially lead to.
11:57
Yeah, yeah, no, thank you, Michael.
12:01
So integration-wise, we see a lot of complexity that can be faced because you've got existing systems.
12:10
We tend with vision systems, you can become more data-driven in isolation, so it doesn't matter the age of the machine.
12:18
By putting a bit of vision on there, you can begin to kind of quantify those decisions.
12:23
But traditionally, you could lean into a complex or a disruptive process when you're trying to integrate that technology.
12:31
Today, there's kind of a few trends that are coming on the horizon, and integration paths are becoming more flexible and accessible.
12:40
So things like modular systems makes a lot of sense, but kind of keep parts of your system distinct.
12:48
And a lot of new solutions, for example, in machine vision, we've got the likes of Cognex or Zebra and kind of those hardware providers.
12:56
You can get smart cameras, And they can slot into existing lines with minimal disruption.
13:02
You don't have to replace an entire system to get the benefits that you might seek.
13:08
We see a lot of work in in cloud connectivity and particularly with machine vision we use edge computing a lot.
13:15
That's so you can make real time decisions, but also enable real time monitoring.
13:19
So, Michael mentioned kind of IT, that's a critical pillar for us when we're thinking about new systems and trying to integrate them effectively.
13:29
We want to tackle those potential blockers up front so that we know what requirements we might need to meet.
13:38
And plug and play sensors and cameras are another thing that are coming to the market.
13:43
So things like Siemens Inspecto, which is an AI off the shelf vision solution.
13:50
But again, thinking about that more broadly, you kind of want to know what problem you're solving with it because all of these different systems have their limitations.
14:01
And sometimes you can enter into automation in one direction and then realize actually it might not integrate the way you think it might.
14:11
So ultimately designing that integration around your workflow is really important and not doing it the other way around.
14:19
One of the other things that we're seeing is customers using kind of commoditized solutions in terms of camera systems.
14:30
So one of Michael's products is replay.
14:33
And what that is is a kind of ring doorbell style error on a machine and then an operator can go and see what that error was and then you can investigate.
14:47
But what we're beginning to do with machine vision particularly is challenge our engineers around using lower cost equipment.
14:54
So we're in an age where you can use smartphones and smartphone processor to do machine vision.
15:01
So instead of looking at that video for problem discovery, we should be able to tell you what the problem is and turn that video into data.
15:09
So we're seeing a lot of kind of lower cost machine vision and applying that in new ways.
15:18
So what used to be considered high investment and specialist, we're trying to make it more accessible in the work that we do.
15:26
So entry level cameras deliver a good performance.
15:28
I'm sure your smartphone, you know, you look at that camera and then you look at the cost of a machine vision system and sometimes bulk at it.
15:37
So we're working on how do we reduce that?
15:39
How do we leverage AI and software so you can configure inspections yourself so we can ship machines that can be programmed by you, but then also there's a balance there because some might say that that might transfer the complexity from us coding it up to you coding it up.
15:57
So it's always finding the right balance.
16:01
And you can apply these solutions to kind of basic presence or absence detection.
16:06
you do counting, we can do barcodes.
16:10
I know an adjacent project where instead of using Cogmex barcode scanning, you can employ a Raspberry Pi, a cheap camera and get the same effect.
16:19
And if you scale that over 2000 barcode scanners, all of a sudden you've kind of, you've made that business case for that saving.
16:28
So what does that ultimately mean when we translate that?
16:32
So there's a lot of quick wins that you can get, but seeing kind of the audience and beginning that automation journey, I think it's interesting to think about how you turn pilots into actual line, full line automation.
16:48
You can scale at the pace you want to.
16:51
And because things are modular and affordable, you can kind of prove value on one line before you roll it out.
16:58
And the underlying trend is data.
17:02
Data is valuable, but data is only valuable in context and when you use it properly.
17:09
So that kind of theme, just thinking about that, Michael, I know there's a lot of kind of internet of things work that you're doing.
17:17
Do you want to go into some detail on that?
17:19
Yeah, sure thing.
17:21
So, you know, you mentioned about modular systems and edge computing.
17:24
So our machines are shipped and fitted with edge devices that we can use and we have standard data interfaces into the PLC because we anticipate that customers require access to this data and a repeat kind of challenge that we saw was customers have a requirement for data and they pass them to us in a URS, a specification, and it kind of becomes a new project every time.
17:52
and with the edge computing technology that we have, we can deploy applications to our machines that sit on top of the PLC network.
18:01
So if that is MPAT Replay, our own machine ring doorbell type system, or if it's a KubeConnect, which we can use to extract data from machines and then provide them to either other IT systems or remote condition monitoring systems, performance monitoring systems, and tie those to human beings that are able to then take that data and act on it.
18:25
And a real shift that we've been able to kind of benefit from and our customers have been able to benefit from is this modular approach to, okay, you want this application deploying to your machine?
18:36
Okay, well, it's well-prepared and the data kind of ecosystem on the machine is ready to accept that challenge effectively.
18:43
And a digital ready, we like to refer to it internally.
18:47
So, yeah, I think certainly, yeah, there's a big link to how, you know, we can extract this data in a modular way.
18:56
Both mentioned there about the quality of the, so data basically is valuable, but only as valuable as the decisions you make from it.
19:03
That's the context that, you know, adds to change in your business, and when we start to think about the commercial impact, which is one of the main subjects of this, so the Hardware costs are coming down, Ed, and all technology levels are getting higher and you can do more.
19:17
Depends how you look at it within the same budget.
19:20
But as you start to obtain more and more data, do you think that the tools such as what Michael mentioned about AI and some of the processing tools, they help you to manage that data a lot simpler?
19:31
Because if you're taking thousands of photos in a high volume manufacturing line, your data acquisition and your data management systems might get blocked up at that point.
19:40
Do you think it is all tied back down to that balance of defining what your problem statement is in the beginning?
19:47
What are you trying to solve so that you can look at the whole ecosystem to bring it all together?
19:54
Yeah, absolutely, Tom.
19:56
It's always a question in machine vision of what do you want to detect?
20:01
So what might look good in a URS where you say I want within half a millimeter this scratch, all of a sudden you get onto a production line and then you turn off your vision system because you get too many rejects.
20:15
And we see this happen quite a lot.
20:19
And it's really understanding, it's one thing to specify that defect that you want to analyze, then it's another once you get it onto the line and see what the reality is.
20:31
Because with vision, we can go down to microns.
20:33
But if you want to find the micron change, then you're going to find a lot of noise in that imagery.
20:40
And then subsequently, you get lots of data out.
20:43
And it's only when you ask, what does it actually, what's important to me that you can really get to using it effectively?
20:54
So what we do day to day is kind of take that human approach of, OK, you maybe don't want all of that data.
21:03
You just want the output.
21:04
So let's just give the output.
21:06
We store the data if it's needed for compliance and quality reasons, but actually at the end of the day, who's operating the machine, what do they need to know and do they need to intervene?
21:19
Particularly around quality, we're seeing a lot of quality systems that previously were the preserve of healthcare customers, but now we're using the same systems but in food and beverage.
21:33
So 360 inspection on kind of source tubes is one of the projects that we're looking at at the moment, where previously this kind of technology wouldn't be applicable because it wouldn't make sense cost wise, but now it does.
21:49
So as long as we're sort of asking the right questions, we can get to the bottom of the data. Perfectly.
22:01
As you can see, it transitions us perfectly onto the next subject, so I thought I'd just change the slide as we go along, which is discussing the role of digital solutions and machine vision and product quality and safety, because ultimately commercial benefits are fantastic, but you need to make sure that your production process is safe and that product quality is there.
22:19
My background is in medical devices, and to what Ed said is that the cost for a recall something because of a print label error is just as much as a product failure.
22:29
So it's the things that sometimes seem like the lowest value or the simplest checks add some of the highest value as they go through. Apologies for interrupting your mid-flow there Azalea.
22:40
I just thought it was a perfect time to segue on to the next bit. Yeah, no, that's very true Tom, thank you.
22:48
And yeah, I mean we're seeing that with the technology, hyperspectral technology is something that previously you might use on farmer type applications because of its high cost but what hyperspectral is, if you're not familiar with it, it captures images in red, green and blue channels and a full spectrum of wavelengths.
23:14
So that means you can pick up the chemical signatures from the image which then means you can detect things that can't be seen.
23:22
That could mean seasoning uniformity, it could mean moisture, and it could mean that you can begin to detect kind of bake quality, for example, in ways that you couldn't previously.
23:37
And when we think about food and beverage, if you go to a vision or an alternation show, you can almost guarantee they'll never show you a crisp packet as part of the vision system because it's really hard to do.
23:50
But again, it's becoming easier with the technology and the tools that we have, which means we can begin to use it in more areas.
24:01
Just touching on the hyperspectral, we're using it at the moment on biscuit lines.
24:07
So we can do surface defects, and then we can do moisture levels and the ingredient mix as well.
24:14
So we're getting down to really fine level details, and particularly in products where you might have same color uniformity, but they have different compositions, you can begin to detect things otherwise unseen in the quality space.
24:34
Also, thinking of that cost, we're also seeing 3D scanning applications, particularly in baked goods at the moment and previously those sorts of pieces of kit with a preserve of car manufacturers, semiconductor manufacturers, but now they're at sensible price points that they can make sense when you look at them in a kind of food background.
25:03
So really all that we're talking about there is a tool to get you to the answer of, Okay, what do I actually want to measure?
25:12
And giving you surety over, okay, is this product really right?
25:16
Is it right to ship?
25:18
But Michael, I know, what themes are you kind of seeing around how quality or detection might influence operator behavior?
25:28
Yeah, it's a good point and a departure from the hyperspectral, but the way the impact replay you mentioned earlier, so our own machine CCTV system, that's used to inform operators about the function of the machine.
25:41
So if we have an issue on the machine, we pop a video to the operator on the HMI to show them what's just gone wrong.
25:49
So a much more lower tech solution, but incredibly high value one nonetheless.
25:56
This is our answer to these machines, automation is becoming more and more complicated.
26:03
The machines are becoming more performance, and there's much more capability out of them, but they're becoming higher tech.
26:10
And one kind of response to that is we want to make sure that the operators are using the machines in a consistent way, in a high quality way, and keeping the machines running as best they possibly can.
26:22
So Replay is our tool to be able to inform the operators, hey, somebody's gone wrong on the machine, you're getting this warning message or this alarm message to some people.
26:33
of certainly very proficient, experienced individuals might immediately know what's wrong.
26:39
But actually we're using it as a more intuitive means to understand the machine and understand how to maintain the machine for consistency and, you know, in some cases, safety.
26:51
You know, just to reiterate, the whole solution is not just about a machine that performs. It's about ensuring that our customers are meeting production targets.
27:00
And that's not just the metal that's kicking out the cartons or the food, it's around how is this machine being used in the organization and how we're maintaining it, and how we make sure that we're getting the most out of it.
27:17
And with that, Ed, I think different materials and automation challenges and how we might help solve some of those is something that you've been looking at?
27:28
Yeah, yeah, I think meeting kind of safety standards is is definitely high on the priority and we look at a number of different applications.
27:42
But providing that level of surety, you know, whether it might be one of your customers that's asking for it.
27:49
We find those situations happen in food and beverage.
27:53
You might have a customer that's not satisfied with the product, and then you have upstream work you might need to do to provide that level of safety or meeting standards.
28:06
Then that's where we might get involved and figure out what kind of vision might work.
28:11
But increasingly, that turnkey vision system that might give you, Okay, this product is uniform all around.
28:21
The seals have been applied correctly.
28:24
Here's the code that sits on it and it's going in this direction.
28:28
Mean that you can pick up those data points that mean you can comply with your safety standards, but you're also kind of leading.
28:36
And often what we see is when you embrace those standards and kind of make them, put them at the front of what you're doing, It can be a real commercial advantage for you when when we sort of have those conversations.
28:56
Suppose we're talking a lot about digital solutions and vision in isolation but to unpack fundamentally at a machine designer which is you come to us with a product, we show you how to make it, package it, palletize it, whatever the processes are, but a lot of these digital solutions and vision applications are needed to simplify the levels of automation.
29:16
to make certain levels of automation more accessible to a new industry.
29:21
As for one, I'm kind of thinking out loud now which is dangerous, but if we look at how we can manage the food industry, so there's lots of stuff that needs to be cleaned down, wiped down, where previously if you had to do mechanical tests on certain baked goods to check the form or whatever it is that you needed to do, that is another mechanical component, another articulation, another actuation.
29:42
Whereas some of the tools that, you know, Ed talked around around hyperspectral imaging or, you know, replay, it allows you to identify product quality a little bit earlier, which means that you can simplify the automation.
29:54
And then replay helps you to identify where there's production related problems as well.
30:00
So it's about the beauty between the mechanical and the physical about what you've got to do.
30:05
And then also how would you maximize that or make the most or simplify it through using those digital solutions and vision applications.
30:14
So what I'd like to do next is to just talk around the poll results, which I'm hoping Luke can pull on the screen now or share what he did before.
30:24
So if we look at all of the votes that we've had through so far, it's clear that the two biggest problems are lack of technical expertise or skills, and then the other one is integration integration with existing systems.
30:40
And actually a lot of stuff that we have talked about almost should retract my previous statement, because a lot about what Ed and Mike have been talking about, a lot of systems can be retrofitted as well, or you can use slight upgrades in order to make more and older systems to be more applicable.
30:58
So we'll try and focus on some of those upgrades and things we go through in the lack of technical expertise or skills does lead us on quite well to the next subject matter that we've got as well, which is examining collaborative strategies.
31:14
Because from the first poll, a lot of people have said that you're on the beginning of your journey, either investigating or you're doing pilot projects, trying to scale, make sure that we can use them.
31:23
So how can you work with suppliers, customers, third-party support, and then also you're going to have multiple internal departments that you need to manage as well, which is a challenge.
31:35
So I suppose, how can we ensure that, or what tools or tips and tricks can you guys provide to make sure that people are aligned and that everyone's pulling in the right direction?
31:50
Go on, I'll pick on Ed to go first.
31:52
Go on, I'll throw you in it.
31:54
Cool, thank you, Tom.
31:57
So yeah, it's really interesting that integration has cropped up because we've had a number of conversations where someone particularly in machine vision might go to market and say well I found the perfect vision system but it stops their boundary is the vision system.
32:18
And what we find a lot of the time is integrated something that we love doing.
32:25
Because it doesn't work unless the system works together.
32:30
and we sort of pull on different specialisms to be able to make that happen.
32:35
So whether we ask the questions of the data, we figure out where the data needs to go and then we plummet in.
32:43
So I think that's a really interesting sort of challenge that's been highlighted.
32:48
But yeah, collaboration is important, right?
32:51
You don't want to enter into something into a new purchase and then be left on your own when things might go wrong.
33:01
So how we approach that with Saiga particularly is kind of a partnership.
33:06
And we tend to offer paid proof of concepts with all of our sort of quotes at the moment.
33:13
And we do that for a number of reasons.
33:17
Number one, when you enter into a purchase of something, it's a risk and how do we balance risk?
33:25
Because ultimately you want to buy confidence, You want to buy confidence that the system will work.
33:30
And for us to do that, we need to invest the time.
33:35
So we can do that proof of concept on the line with real data, not kind of a setup in a lab.
33:43
It tends to be a focused project and we kind of work alongside customers to solve that problem for you or work on the production line or use your actual products under real conditions.
33:56
And where we found that really helpful is also in business case.
34:00
So not only are you saying, hey, look, we've got a vision system that we think will do X, Y, Z, you can say, well, you know, we've had it in, here's the results, fundamentally we can increase our efficiency or our quality by X percentage, therefore, you know, we should make that investment.
34:20
So it's really all about lowering your risk and collaborating to do that.
34:26
It's about evidence. So again, coming back to the data, but it is fundamental kind of we, we generate that together.
34:35
And then we work collaboratively.
34:37
So working with with the people who are using the system, with the engineering managers and making sure that actually those processes that you have in place today, we might uncover that we need to factor in some additional work to meet those.
34:57
But really, it's about how do we collaborate to find the right solution.
35:03
and it's doing that in partnership.
35:06
I don't know, Michael, what have you seen work here?
35:09
Yeah, I mean, just on collaboration, actually something that I've been working on very recently, which was surprising.
35:14
So working with one of our customers that has been a long-standing customer, but not actually ordered a machine in something like 20 plus years.
35:21
So when you imagine those skills actually within the business of how do we order a machine and how do we describe exactly what we want, working with that customer at that level to say, right, we can help you develop a URS, which is an interesting one, but then, you know, helping them navigate where the change, the recent changes are in cybersecurity, you know, when you're looking at NIST 2 and helping them understand, okay, how does that change the way that the HMI needs to be designed and how, you know, operators are able to interact with it and what other IT systems the HMI can integrate with and just being able to hold their hand through that process and, you know, draw on the experience that we have to help them sort of make sense of it all because it can be incredibly daunting at that stage of, we want to buy a new piece of automation or we want to buy a new piece of integration.
36:08
It's, where do we start?
36:10
We are literally starting at the beginning with them and holding down through it.
36:17
What are some of the ways that you might enter, Ed, with a customer and sort of start things off with them?
36:25
Yeah, yeah.
36:26
I mean, like I said, we've had a number of conversations around, yes, I might have found a vision system, but it doesn't work with our equipment.
36:38
Or actually, it's all well and good to say you can do a modular system, but what happens when the piece of kit's 10 years old and it's difficult to work with?
36:50
Those are the kind of conversations that we have quite a lot.
36:56
But really, just bringing it back kind of where we started this, it's about defining that problem effectively.
37:06
So further down the funnel, when you've decided that you want to go ahead and automate, I'll often try and ask the question, okay, well, what's the purpose?
37:17
Or even what's the company's strategic goal over the next three to five years?
37:23
Because if it's consolidation, and you're asking for a system that's going to put your throughput through the roof, it's understanding the context behind that.
37:35
Why are we going for high numbers when actually a more pragmatic approach might be beneficial for certain systems?
37:46
So really, the best way to kick off kind of that collaboration is going a bit further back to ask that original question of what are we actually trying to accomplish here so that we can make sure any solution or any suggestion is sensible and grounded in, why do we actually want to do this?
38:06
And that's always just through conversations.
38:09
It doesn't cost anything to have that conversation and ask those questions.
38:13
And I see that as a key part to working with customers in this industry is, well, if I can further the thinking or make you ask those questions yourself, then I'm helping.
38:27
because it's really important that we have those answers before we start on this journey.
38:33
I don't know, Tom, have you got any thoughts on that?
38:36
Well, yeah, just on a couple of things you said there, and if we look at the last poll that we had is that, you know, a third of the people, lack of technical expertise or skills, their biggest challenge to adopt.
38:48
It's now, the way in which we've structured as a group is we've got multiple different brands that work there, but Ed and Michael don't sit within one of those brands.
38:56
they sit, you know, they cross the whole spectrum because we want to leverage our best expertise internally as much as we can.
39:04
So, Ed could be pulled on one day to look at a packaging requirement for a palletiser or an orientation or there may be a vision quality thing.
39:13
The next thing you might be doing is looking at a medical implant, implanted device that's where microns matter.
39:19
So, when we were saying earlier, when do they matter and when they don't?
39:22
There is going to be applications where they do.
39:25
Now, what Ed and Michael, and there's many people in our business that are like this, they see lots of different things.
39:32
They can pick and choose from their experience what the right aspect is.
39:36
And I suppose, particularly for those that are earlier on in their journey, don't be afraid to engage with supply chain to help you define what that problem statement is and to define what that proper one is.
39:47
Because and we're not unique in that we have people that work across the board but you know engaging in a conversation and talking to your suppliers or partners or people maybe that you're not too familiar with that you want to reach out to have a conversation because it does it helps you to clarify your own thoughts.
40:06
I sat in a couple of meetings where Ed's been talking to people internally and to customers and asking a few simple questions that actually when you're just considering the problem statement seem really obvious to ask, but when you're caught up in the weeds of I need to produce this many products, I need to get it, you know, this many metrics, you know, I've got this budget, you often forget the reason behind what you're trying to investigate.
40:29
So, yes, my thought would be just don't be afraid to reach out to people and talk around some of those things about what's, you know, what's the key things you're trying to achieve, but actually is this the right thing that we should be looking out or maybe for instance turn around and say well you could do that or we could apply this technology which is a third of the price but gets you 90% of the way there and it's it's about managing what what's the art of the possible and you might have people internally as well that you haven't aren't aware of or haven't reached out and so yeah we would we try and advertise internally what people are doing so that everyone knows you to reach out to but maybe people in your own business as well but the easiest thing I believe in first instance is to start talking to people about it, what are the challenges are.
41:16
And I suppose, Ed, you're looking from a vision perspective and a product perspective, but from a digital solution, is there anything you can think of, Michael, that, you know, first-time customers and we're all in the UK, this is the Food and Drink Federation of the UK, so it's not too bad, however, there is an element of how can you span the people that live in the Scottish Highlands or down on the South Coast?
41:41
Yeah, so another product that I'm working on, Keep Connect, allows our service team really to connect with the product with the automation.
41:51
And what we're finding really for first time customers, this is a vital tool that allows us to connect with them to make sure that if we sell them automation and if we design a product around their requirements, we can ensure that it does the thing that it's designed to do and give them the support that they need to ramp up.
42:09
Because there is a handover period, you know.
42:11
We say, okay, there's some new automation, there's a new system here.
42:14
You're kind of, you're running this now.
42:17
You know, we've trained you and we've given you the tools that you need, but in, you know, having the tools to be able to keep an eye on the production and make sure, okay, everything's going smoothly and you're ramping up in the way that we'd expect to see you ramp up.
42:32
And we're finding, you know, specifically with first-time customers that that you know that remote service and that remote monitoring is is vital to make sure that they are where they're supposed to be.
42:46
Thank you and and what I was planning on doing now is to just sort of run over the questions because we've received a few as the time has gone on on various different subjects so we'll try and work our way through but Ed or Michael have you got anything you'd like to add just as a closing statement or are you happy to move on to questions?
43:06
I'm happy to see what the questions are. Perfect. We've got a few that have popped in.
43:13
Now we're at a quarter to the hour and we've left a significant chunk to answer some of the questions.
43:18
We've got a few through as it stands.
43:22
Let me pick out, so if we start with one that's a little bit more general, so lots of our data sits in dashboards but isn't used effectively, how would you suggest we bridge insight and action?
43:34
And I'll open that up to either of you, to be honest.
43:36
That could be one for both of you.
43:38
Yeah, I mean, that's it's quite relevant to something that I'm working at the moment, which is odd.
43:43
But, you know, one of the things that I like to do is find that that champion, that cluster of people.
43:48
And, you know, it might often not be a bad idea to start small and gradually bring people in.
43:53
But yeah, I mean, we're working with a customer that has access to this data and they have done for a while.
43:59
And really it can be quite hard for different parties to interpret.
44:03
And what I'm doing at the moment is identifying who is that champion?
44:07
Who's the person that can link those systems together?
44:09
And how do we provide them with the resources that they need?
44:12
You know, it's all around identifying the person that is excited about it and then making sure that they are heard and responded to.
44:19
So as a product manager, you know, it's really the lifeblood of what I'm doing is bringing in requirements and making sure that I can turn those requirements into a function that those people can use, and then really being able to give them the tools in a really rapid way.
44:34
So they say, right, I've got this problem.
44:36
That's okay.
44:36
How are we going to solve this?
44:37
Okay, try this.
44:38
Try this new system and make sure that as they're demanding more from our IT systems that we're able to provide that to them.
44:46
And you get some pretty phenomenal acceleration that way.
44:49
It's, yeah, it's quite interesting.
44:51
And then, if we can get them enthusiastic and excited about it, it does become infectious.
44:55
And then they start to bring more and more people into the fold.
45:02
Perfect sounds good and then another question we've got which is a little bit more specific but have you done any automation in seafood aquaculture products and processing factories so and I can talk a little bit about some BCA which is Boston Conveyor and Automation over in the East Coast of the US those guys have done multiple products with frozen fish foods manipulating handling and food processing those and again these incorporate vision systems, because if there's anything about grown products, fish, vegetable, anything like that, is they're never the same.
45:35
So there is a lot of vision that's required from that perspective.
45:39
So we have experience handling, particularly the frozen products, but Ed, is there anything that you've done within this sector or something you can draw parallels from, from a vision perspective in those industries?
45:50
Yeah, it's a good point Tom, around kind of the variability in the data.
45:58
So what's just looking around seafood?
46:01
What makes good seafood?
46:02
What makes bad seafood?
46:03
How do you measure it at the moment?
46:06
And where technology is really advanced is being able to train those models in the way that you might want them trained personally on your shop floor.
46:17
So where you get that variability, no longer do you have to put boxes on and do areas of interest for machine vision.
46:25
you can give a picture and say well actually that's good and tell me when it's not good or tell me certain aspects about this image and how it changed.
46:36
We have looked at doing in an adjacent industry kind of accounts for mosquitoes and kind of being able to do that with vision systems, being able to provide confidence metrics around kind of what we're detecting.
46:55
So, yeah, we've seen a few different applications, but it's really around where variability is used to be an issue.
47:03
Now, it doesn't have to be so much of a big issue in these kind of applications.
47:12
Thank you, Ed. Appreciate that. We've got a couple more that have come through.
47:17
So, one's I'm struggling to scale digital machine vision beyond our pilot project.
47:23
How can I make it I suppose that's applicable across multiple industries.
47:29
It might be the same as the question from seafood.
47:32
It could be used across any, but have you guys got any advice?
47:40
Yeah, I'll go first, but you're not escaping this one, Michael, you will need to answer it.
47:46
My perspective is, and I think this kind of was born out in the polls.
47:53
How do you make it easier or sound like a broken record, But a lot of the time, people, when faced with challenges, will ask more questions.
48:04
So it's kind of having those answers in place before the questions are asked.
48:11
So we experience quite a lot of friction from an IT perspective.
48:17
Or there might be difficulty that you might have a certain vision supply, or you might have a certain supplier of sensors.
48:24
And actually, you should probably go to them.
48:26
But the machine vision market is crazy complex because you've got certain technologies and vendors who don't want to work with end customers and then we kind of sit in that middle realm.
48:40
But when you get sold a system, you don't want just a system, you want a partnership, you want to figure out if you're actually trying to change your business process.
48:50
You don't just want hardware, you want kind of more.
48:54
And I think it's a really interesting point because if you define that requirement precisely then you can look at different options and make things easier.
49:07
So at the moment we're doing low-cost vision for anomaly detection of boxes going into palatizers.
49:16
Pretty simple use case, but then we've got an adjacent use case in healthcare of, can you inspect this cotton and make sure that there's no dents or tears on it?
49:28
Where normally I spec a high quality camera system with certain software to do that.
49:35
We've developed the software for anomaly detection, so tears, rips, corner dents, that kind of thing.
49:43
Why can't we just apply it there?
49:45
So it's kind of defining that again what are we trying to solve and then step by step bringing people on that journey.
49:55
So it's generally around communication and thinking ahead to all the questions that you'll be asked when the time comes. I don't know Michael what are your thoughts?
50:06
Yeah I mean from a digital systems perspective thank you for going first Ed by the way you've obviously given me a chance to ever think about that which is great.
50:13
From a digital this perspective, the way I approach this myself, and this is not necessarily the best way, but I've had some successes.
50:21
You know, we've got to an MVP point.
50:23
You know, if we're saying we're struggling to advance past pilot stage, the assumption is that there is a pilot there.
50:31
My kind of playbook in that sense is always what is the bare minimum effort I can put in to get me to the next threshold?
50:40
You know, how do we just add that little bit on that's gonna provide more value?
50:44
What's the smallest iteration I could go through to produce an extra little bit of value?
50:50
And one thing is that allows you not only to see, okay, I'm producing more value from my initial pilot, but also to measure that we're going in the right direction.
50:59
Something can be quite difficult.
51:01
It can be really, what's the word?
51:04
It can be really enticing to say, oh, right, we've got an MVP here.
51:07
This is great.
51:08
This can be fantastic.
51:09
And here are all the things that we can do with this and here are all the problems that we can solve with this.
51:14
And, you know, boiling the ocean is an age old adage, but for me, it's how do we iterate the smallest possible increment whilst investing the least possible amount of effort and then show the increase in value and test it and make sure that we're going in the right direction.
51:32
Perfect.
51:33
Now I'm gonna really try and catch my clout so I'm gonna give you no time to prepare for this one.
51:39
So one of the questions is how do we align internally on our problem statement?
51:42
something that's come up numerous times on this call but what's some advice for managing things internally?
51:48
Yeah I mean something recently I've done again I mentioned right the start capturing user stories so if we've got multiple stakeholders that all need something out of a problem you there is you know obviously systematic way that you can approach that and it could be as simple as okay Tom I'm going to ask what you want from this system and I'm going to ask what you want from this system I'm going to write down your user stories you know Tom wants this because he's to achieve that.
52:12
And then it's just, you know, it can be as simple as performing a Moscow analysis on those results.
52:18
You know, I've got a list of things that I want my new digital system to do, a list of people that it needs to satisfy.
52:27
We can say, okay, it needs to have that, might not need to have that, and you can do that as a group, as a group exercise.
52:34
And actually, what's quite interesting is, once you start doing these exercises, so much more value comes from just having those discussions with different stakeholders in the room.
52:44
It's like, oh, wow, I didn't realize that you valued that as important as you do.
52:49
Let's go into a discussion about that.
52:51
Well, yeah, I need this data from this system because it's gonna enable this team to save two hours a week or four hours a week.
52:57
And then you start to get into those discussions.
52:59
Then as a group, you can agree on, okay, well, the new system that we're targeting has to have this.
53:06
And these are all the things that are really important and why, and then that can be documented.
53:10
And again, that also allows you to, as I was saying before, your tiny increment that you wanna do to show some value out of the system, you can say, well, our tiny increment is gonna be this because we can get so much value from it.
53:22
And it might not be the top most valuable thing because you might decide, okay, the most valuable thing that we can do is actually gonna cost us 10 times more development power, but actually we can do the 80th percentile, you know, 20% of the effort.
53:36
So, you know, doing a good old user stories and Moscow analysis is something that I've done in the past and had a lot of success from.
53:46
Well answered, given.
53:48
I thought I was going to catch you off guard there.
53:52
And the next question we've got is definitely an Ed one, sorry Ed, but it's, you talked about quick wins with low cost vision, finding the right balance, what's a good example?
54:02
I'm not sure what industry the question is from, but I'll let you with free rein.
54:08
Yeah, yeah, quick wins.
54:12
So we're finding traditional sort of mechanical or detection methods like infrared or things like that.
54:24
You can begin to replace them quite quickly with kind of our low cost vision systems.
54:29
So the most applicable one would be the kind of anomaly detection that we're working on.
54:36
So at the moment, what's used is infrared sensors that can detect if a box flap is open.
54:44
So for the same price point, what we're developing is a system that does flap detection, box damage, corner damage, and all those sorts of things that could mean a pallet when you send it out of the warehouse falls over.
54:59
So you can begin to pick that up.
55:02
So we're seeing a lot of systems where, for example, in pallet check as well, we're looking to compete with an advanced optical pallet check system with the same price as it costs for a mechanical check.
55:16
So that age of mechanical driven systems is beginning to sort of lessen and actual vision data can be used.
55:25
and so the the clearest example is we've got a few relatively low cost cameras we can put machine vision over the top of it and begin to get data and outputs out of that and we're actively doing that today with with our palletizing systems.
55:45
Perfect then that actually brings us to a close of the questions which actually works out quite well from a timing perspective unless anyone wants to ask anything very quickly.
55:55
I think that draws this to an end.
55:58
Yeah, thank you for those that have dialed in as well.
56:01
And thank you again for those that have asked questions, been really useful.
56:05
So what we will be doing as well is, I believe Luke will be following out for those that have attended with an email, just a bit of an overview and some contact details if you wanted to get in touch with us for any specific questions afterwards or once you've had some time to think.
56:20
So I suppose from me and the rest of the panel, thank you very much for attending today. It's been a good conversation.
56:27
Thank you. Thank you very much. Thank you to you two guys as well.
56:31
And behind the scenes we've had Donna who's been supporting on this from our marketing department as well.
56:35
So thank you very Thank you very much.
