IRI Industrial Robots Insight

IRI is an integral, easy-to-deploy solution to turn industrial robot behaviour data into value through predictive maintenance

Why choose IRI?

The product



IRI is an integral, easy-to-deploy solution to turn industrial robot behaviour data into value: Enhance performance, reduce energy expenditure, extend machine lifecycles, improve reliability and product quality, that is, it allows the predictive maintenance of the robots.

A big data product taylored to process the data of any KUKA or FANUC robotic arm connected to SIEMENS PLCs extracted through various industrial protocols (Profibus, Profinet, OPC UP, MTConnect…), which generates real-time predictions and alarms taking advantage of deep learning algorithms. This will allow the plant worker to make decisions. Moreover, IRI will enable the client to study process quality by means of statistical analysis of the robots' behaviour.






IRI focuses on the communication of both Industrial and Big Data realms through some of the state-of-the-art protocols, i.e S7 Profibus and Profinet. Data is input into the cluster, formatted and pre-processed, and stored according to a data model suitable for addressing the challenge needs, with a time-series orientation for managing robot axes data nature, supporting up to the millisecond granularity. Getting several robot behaviour cycles, IRI trains the time series by means of the Long Short-Term Memory (LSTM) method in order to predict the subsequent robot performance. This is crucial to make sure that the robot is behaving properly by comparing the forecasted data to the real-time data, and hence, to predict failures in the robots. This leads to early alarm detection and error failure prevention. Furthermore, implementing advanced analytic techniques and taking advantage of the historical data, IRI will be capable to make a deep statistical analysis of the processes being monitored and estimate the quality, efficiency and availability.


Final users


Data visualization is oriented to different levels of abstraction, ranging from low to high, where lower level is characterized by simple variable monitoring and instant robot behaviour, medium level builds over previous information towards alarm detection and robot performance, and higher level adds further concepts for creating different perspectives of data. These levels could offer interesting data visualization to the different user profiles within the company, from Plan Operators to CRO, CTO, QoS Department and COO.