How to nurse a sick robot back to health

By and | 5 minute read | April 25, 2017

Hannover Messe 2017 series image

How do you know when a robot is feeling under the weather? You can’t tell when a robot is not well just by looking at it – you need to use data to help understand what’s happening.

Today, for example, if I imagine all the robots on the floor at Hannover Messe are on an assembly line – how am I to know if they are in top shape? It’s a bit warm in the hall – from the machines and the people – warmer than it was last night when I left.  If I worked as a line manager for a manufacturer, I’d want to know exactly how each of these extraordinary machines are feeling – are they overheating, suffering from stress, are their parts clicking when they shouldn’t be?

To learn more, I caught up with Shane Philips from IBM at Hannover Messe in the IBM Booth to talk about robots and how they communicate with their human counterparts to tell us when they’re feeling poorly. Here’s what Shane Philips had to say.

Machines tell us how they feel through data

A robot or machine tells us how it’s feeling through data related to its movements, parts, temperature, pressure, in addition to other data sources  – for example, maintenance data from IBM Maximo, or weather data coming from IBM Weather.

The combination of these different data points tells us how that robot is feeling at any given moment. Access to this data allows us to give the robot a health grade – enabling us to apply the right prescription to that robot to get it well. Using a set of Watson algorithms, we know a robot will feel bad when it starts slowing down, if it gets overheated, or, if it vibrates in a certain way. Alternatively, there may just be a set of digits that are spit out in a certain way – to indicate that something has gone awry. This is how a robot tells us it’s not feeling great.

Giving machines an asset health grade

At the booth, Shane Philips walked me through a demonstration of Maximo Asset Health Insights. On his laptop screen, I could see a list of queries pertaining to a set of critical assets all of which had something in common, and which need to be tracked.

Figure 1: Maximo Asset Health Insight – example queries

In this instance, we are looking at data flowing from a set of pumps.  According to the dashboard on the screen, we can see that some of the pumps are feeling fair, but none are feeling poor. The objective is to get them all to feeling good.

At present, the pumps are receiving a health grade based off sensored and monitored structured data, IBM Maximo maintenance and operations history, how long they’ve been running, up time, any maintenance the pumps have received over the years – which is critical as the pumps could be either an aging or young asset. The dashboard is also displaying weather data – but if could be any external source of data – that makes sense to add into an analytic formula. The components of the health grade dashboard are based off of reliability and maintenance best practices.

Getting into a predictive mind-set

Arc Research recently found that as much as 50% of preventive maintenance work could be eliminated by shifting to a condition-based maintenance cycle. Understanding how data such as age, service history, failure history, and configuration changes over time can provide opportunities to significantly improve preventive maintenance efforts.

The objective for any machine or asset is to run smoothly, 24 x 7. A predictive mind-set puts a manufacturer ahead of the curve when it comes to planned downtime. Using predictive analytics capabilities allows an organization to gauge how many times an asset is touched, or taken down, based off what has been prescribed for it.

IBM Maximo Asset Health Insights enables organizations to take advantage of predictive analytics. In the example we talked about earlier, the health grade can tell us when a robot, machine or other critical asset is starting to feel bad. What that means is that the robot is scheduled to have a check-up, or, it will trigger something that tells us it’s going to fail, leading to a reactionary, unplanned situation.

Figure 2: Understanding current conditions alone is not enough

Condition-based maintenance insights to detect when ‘like’ assets are behaving differently

Having condition-based capabilities is especially important where there are a quantity of assets (or robots) which look exactly the same – they are the same model after all. But just because they are alike, doesn’t mean they all behave exactly the same way – especially when it comes to the conditions in which they are operating. For example, if an asset is performing its duties in a desert environment, versus a wetter environment such as Oregon, weather will be a factor that impacts how that robot or asset feels and functions. Even when the machine is the same model, carrying out the same task when the conditions of its environment vary means the machine’s health will vary too.

Creating a baseline scale is the first step to becoming a cognitive business

Downtime for a production line robot is something to avoid. Maximo Asset Health Insights allows organizations to create a baseline scale to assess the status of an asset. By monitoring asset factors, such as condition-based maintenance, asset costs, and performance, it is easy to detect when an asset is failing and subsequently enabling its operators to make improvements quickly.

Using analytics is the first step to becoming a cognitive business. Maximo Asset Health Insights offers manufacturers an easy on ramp to the advantages of cognitive manufacturing. The solution offers organizations a robust set of algorithms available as a template. For industry specific scenarios, there is a workbench that enables an organization to take what’s available out of the box, and customize it easily so their health check aligns to the unique conditions of their critical asset requirements.

Learn more:

  • Read the report, Using the IoT for Preventive Maintenance, to understand how data such as age, service history, failure history, and configuration changes over time can provide opportunities to significantly improve your preventive maintenance efforts.
  • For more information about IBM Watson IoT for Manufacturing, please visit our website.
  • To learn more about IBM Watson IoT, please click here.

You can keep up to date by signing up for Operational Genius, a monthly Watson IoT newsletter designed to be your one-stop resource covering news about IoT and analytics, predictive maintenance, facilities management, asset management, intelligent machines, and more – delivered direct to your inbox.