Industrial robots are an indispensable asset in manufacturing and assembly lines. Their endurance, speed and precision make them suitable for highly repeatable functions that demand accuracy, such as welding, painting, packaging and palletizing.

Robots are particularly important for automotive manufacturers. Imagine a line that produces a car every minute. If a robot that feeds painted panels to the final assembly line suddenly fails, the line starts losing revenue to the beat of one car per minute.

If maintenance personnel take 20 minutes to identify the problem, fix it and commission the robot back to production, that one failure alone will have set the assembly line back by 20 cars, or roughly $400,000 in lost opportunity cost.

There are several reasons why robots need to be maintained. Faults, soft failures and hard failures are all problems that can lead to losses in productivity, so it is important to understand what these events truly mean.

Faults: A fault is an anomalous event that happens due to an incorrect signal input or decision within the system. The impact on production is low, though still troublesome, and the robot health status is generally normal.

Soft Failures: Soft failures are the product of the degradation of robotic system components. They do not result in downtime, but still affect performance by causing diminished throughput (longer takt time, lower yield rate and/or position accuracy loss).

These failures are concerning because, if a weld isn’t corrected, it may lead to low product quality, scrap or, in worst-case scenarios, hard failures.

Hard Failures: Hard failures are often the outcome of a broken component or modules that result in downtime or equipment shutdown. This is the worst kind of failure, mainly because it renders the robot unable to produce any parts.

The aforementioned faults and failures often lead to diminished production yield accuracy caused by variations to robot parameters such as velocity, force and torque.These parameters can be used by data-driven predictive systems to detect incipient fault indicators and infer how fast they are changing over time, so hard failures can be ultimately predicted and prevented.

The practicality of such an approach lies in the non-invasive nature of monitoring applications, since robot parameters are already inherently available for the robot’s control and feedback system.

The primary challenge is whether or not the robot manufacturer enables the collection of these parameters. Assuming robot application plug-ins (APIs) are available, predictive software platforms such as Predictronics’ Factory Sentinel for Industrial Robots can be utilized to analyze parameters with a multi-variate and multi-regime approach that results in robust and accurate robot health calculations.

If you would like to learn more about robot failures and how to predict and prevent them, contact us here.