Raven Sense™: Considering an AI User Experience for Manufacturing

Digital interfaces are becoming more common in increasingly hostile manufacturing environment. Noise, dust, and hazards all conspire to make useful interfaces very difficult to design. More importantly, the expectations of designers and the reality of the shop floor are often at odds. Add to this, machine interfaces that are increasingly complicated and other sub-systems that demand constant operator attention. The manufacturing shop floor is a challenging environment that needs an elegant, thoughtful and discreet user experience.

What is Raven?

Raven’s mission is to make manufacturing better by using artificial intelligence. 
They help manufacturers leverage all their data. Their first product was very similar to performance management software available today with performance dashboards and stock reports. Their “a-ha” moment came when they realized that the novelty of pretty reports and dashboards wears off and the realization that un-interpreted data provides little value. What manufacturers need is guidance not reports. This realization has guided Raven ever since.

The Raven ecosystem provides manufacturers with a holistic solution to improve operator performance in real-time.

Raven Sense™ connects machines, processes and operators collecting critical performance data like uptime, job, down-time code, production rate, yield and much more.

Raven Sense

Raven Brain™ uses machine learning to learn from the client’s data and processes helping them truly understand what drives their performance. Custom analysis from a dedicated team of data scientists, along with proprietary technology, partners with them to drive transformational improvement.

Raven Flight™ makes sure the right person gets the right information at the right time. Real-time event escalation and opportunity detection with Raven’s mobile app gives your leaders the information they need to enable their teams to perform at the highest level.

Raven Flight

Raven Sense™ is the first part of the ecosystem to be placed in trials with Raven lead customers.

The Challenges

Raven Sense™ provides a first line of feedback to the Raven Ecosystem. Machines themselves provide limited data to manufacturing performance systems and often way too late to be useful. Raven Sense™ is an interface for operators to supplement and contextualize events during the manufacturing process. More importantly, it provides realtime feedback with respect to their performance, quotas and even anticipated deleterious events like parts shortages or sub systems failures.

The environment for Raven Sense™ is often hostile and fraught with distractions. Extreme noise, dust, oil and a variety of other environmental hazards make a useful digital interface very challenging. Operators are required to conform to workplace safety regulations that may require hearing protection, eye protection and clothing that provide an additional layer of isolation from any digital interface.

Operators themselves are typically unaware of their performance and may not get feedback until their end of shift or even days later. More frustrating is their inability to communicate quickly or effectively when their performance is compromised or there is a system failure.

Complicated or involved interfaces are often a barrier for manual labourers and will not be easily adopted. The baseline for acceptable complexity with this demographic is usually something akin to a mobile phone interface or an ATM kiosk type experience.

The Solution

Relaying Performance State to Operators

Raven Sense™needs to relay performance health to the operator. This provides them with the reassurance that “all is well” or that they need to take some action. An additional complication may be that a single operator is responsible for multiple machines in an area. They need an affordance that provides a “single glance” as to the machine state.
This challenge was addressed by modulating the screen color;

Green:
Running at a good rate (active). The green color represents a condition of normal operation. The device should always be green as it shows that there are no blockers and the process is performing as planned.

Red:
User input required immediately The red color represents a condition of inactivity. The device should never be red. When red, the operator must interact with the device to provide additional context. Once completed, the device state will change to a yellow color.

Yellow: 
Suboptimal Performance. The yellow color represents a condition of suboptimal function. If yellow, something has happened and therefore the process is operating at a suboptimal rate. The operator is not required to do anything at this point, the screen is simply indicating that performance is below rate for this shift.

Operator Input

Given the challenging environment on the shop floor, a simple input pattern was conceived for operators. Noise is a distraction in most manufacturing environments, speech inputs were not an option. The design guidelines limited the operator to simple touch and select operations, eschewing any keyboard or mouse interactions.

Tagging

Tagging is a way for operators to bookmark or respond to milestones during the manufacturing process. It provides critical human feedback to the data collection to clearly indicate what is happening at this specific machine. When the interface displays a “red” state (operator input required), they simply touch the screen and are presented with a scrolling list of choices. These will vary from customer to customer but can generally be summarize into a dozen of simple response statements (malfunction, parts shortage, shutdown, etc). The operator selects the appropriate response, is provided with a reassuring (non-modal) confirmation and the state of the device is changed. More importantly, the operator can also pre-emptively provide an input (in the green state) if they anticipate an issue or condition.

Escalation

A much greater concern on the shop floor is a timely response to incidents that will halt production. Machine malfunctions, parts shortages and other events beyond the control of the operator need to be flagged immediately and appropriately. Leveraging the same pattern as “tagging”, the operator can start the escalation process by selecting the “Call” button ever present on the main screen. Some installations will provide a second level of call direction (supervisor, lead hand, etc) to allow a more targeted reaction. Once selected, the escalation process is started and a chain of events is triggered through the Raven ecosystem.

Operator Feedback

How am I doing?

An important aspect of the device is to provide real time, positive feedback to the operator during their shift. At any time they can assess their current performance and react accordingly. In addition to the The Raven Sense™ main interface colors, the operator is provided with relevant performance information on the main screen. These a simple number tallies that are embodied differently in each unique customer location. Essentially, the screen will display a current count of production, relative to the target count. It may also display a velocity, indicating the likelihood of achieving the quota. Some installations may display an OEE (Operational Equipment Effectiveness) as a measure of the operator’s efficiency.

Positive Feedback

The baseline information provided by the device and the operator is fed back into the Raven ecosystem and a series of realtime reports are sent back up the chain of command. An automated and discreet toast interface provides feedback to the operator as they meet or exceed performance expectations. Additionally, any of the supervisors, managers or even the plant manager can “like” or personally congratulate the operator in real time.

Raven Sense™ is currently in a trial phase with important lead customers and the feedback has been positive. Real world testing has revealed some interesting edge cases that will require another iteration of the design. More importantly, the team is moving forward with the other critical pieces of the Raven ecosystem to transform manufacturing for the better by bringing artificial intelligence to the shop floor.

Like what you read? Give Kevin Deevey a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.