The robotics industry has emerged as a key focus area for Google in recent years. The company’s Boston Dynamics subsidiary recently unveiled a four-legged walker that can independently navigate around a house, while its artificial intelligence researchers are working to improve the finesse of mechanical limbs. And now a new patent application indicates that yet another item has been added to the agenda: The industrial robots used on production lines.
Google’s filing contains the specifications for a cloud-based service designed to tap the diagnostics sensors that are included in most modern factory automation equipment. They’re usually small transmitters that provide updates about a robot’s performance patterns, such as its power consumption levels and the amount of pressure placed on important components. Factory operators route this information to a backend server cluster where it can be examined by maintenance personnel to find potential issues. But picking out important details the old fashioned way is becoming increasingly difficult due to the volume and complexity of the data generated during production.
As a result, industrial companies are adopting various specialized tools for analyzing machine-generated logs, which is what Google appears to be interested in providing. The service outlined in its patent filing uses machine learning algorithms to quickly ingest equipment logs and identify anomalies. If, for instance, one of the motors in a car assembly arm experiences dangerously high torque during an operation, the service will send an automated message to a technician asking them to take a look. Google says that such notifications can even provide specific repair instructions in case there’s enough diagnostics data for its algorithms to identify the cause of a problem.
Moreover, the company’s patent filing indicates that alerting conditions will be continuously adjusted based on new data from the factory floor in order to improve detection accuracy. This functionality is designed to reduce the risk of the service generating false alerts, while at the same time improving its ability to spot subtle problem indicators that reflect real issues. The idea is to help organizations identify and remedy problems in their robotic equipment before they start seriously impacting production.
If the service ever leaves the drawing board, Google will be in a much better position to court companies from the manufacturing and retail sectors. But the search giant can expect fierce competition from the other vendors that also are offering solutions for processing machine-generated data. IBM, for instance, launched a set of machine learning services late last year that can analyze logs from heavy equipment to produce automated maintenance alerts. And industrial giants like General Electric are also developing similar offerings for their clients. Google’s entry into the race is poised to make this already crowded market even more competitive than it is now.