Eliko offers a precise indoor positioning system as well as applications. While it fundamentally provides location information on tracked objects, we can expand the system with more insights. Additionally, we can define paths and movement patterns, estimate speed and measure the duration of events and processes. The data enables us to build applications and analyse and solve problems in various industrial settings. For example, we can optimise the locations of assets, paths or layouts. All of this requires data analysis and knowledge of the details of the underlying processes.
Machine learning algorithms can be applied to improve RTLS performance and accuracy
Challenge: When you use RTLS technology in an empty room, it is easy to measure the correct location coordinates. In the real world, however, there are often obstructions in the tracking area. Walls, people, machines and other assets have an effect on the accuracy of the location coordinates.
Solution: We can use machine learning algorithms to improve the reliability and accuracy of the measurements. By collecting data we can train algorithms to detect and classify different LOS (line-of-sight) and non-LOS situations. This allows us to discard, weigh or even correct measurements according to the classification. As a result, in complex environments where line-of-sight (LOS) measurements are not always possible, the machine learning algorithm mitigates inaccurate measurements and, therefore, improves system performance.
Optimising manufacturing processes by matching patterns and predicting behaviour
Challenge: One of the main ideas behind Industry 4.0 is to be more flexible and efficient in manufacturing. This requires knowledge of the details of the underlying processes. Therefore, there are processes which need real-time tracking to provide accurate data to decision-makers. However, it is a challenge to analyse and read the collected data for decision-making.
Solution: Based on the location data the system generates, we can train machine-learning algorithms to match patterns and predict behaviours. Thus, the second application is on the data side. For example, by analysing the paths and frequency of accessing manufacturing materials, the system can (in real time) suggest optimal locations for storing them in manufacturing or the warehouse. This would help manufacturing workers reduce the time they spend on finding and accessing a specific material roll or pallet.
In conclusion, coordinate error corrections on the one hand and output data analysis on the other are the two main reasons machine learning is beneficial in an indoor location system. With the help of these machine learning algorithms, it is possible to give an important boost to manufacturing efficiencies in Industry 4.0.