Learning by locating: Digital transformation of logistics – Revista Digital


A buildup of containers has caused a jam at deepwater ports such as Felixstowe in eastern England – Copyright AFP/File Philip FONG

New research has shown how artificial intelligence can be used to improve the location of objects within industrial environments. This involves the use of deep learning algorithms to increase the accuracy and performance of current systems.

The reason this research is of interest is because indoor positioning technologies are a driving force. behind the digital transformation of the industrial sector. This is based on the ability to accurately and cheaply track objects, assets and people. The intention is to improve accuracy and save resources, time and money. The types of sectors investing in such technology include logistics and healthcare.

With the new study, to evaluate the potential of different systems, researchers from the Open University of Catalonia will carry out an experiment in a 1,000-square-meter facility. Here, indoor asset location can use the point of view of a receiving device to deduce which direction a signal emitted by objects is coming from. The technology that has been developed has been called ‘DUNE’.

DUNE can translate the collected information into an estimate of your position. Accuracy has been improved through multipath propagation, a process that causes radio signals to arrive at receiving antennas along two or more paths and at different times.

To fine-tune the positioning mechanisms, DUNE is assisted by deep learning techniques at various stages of the positioning process for optimal performance. With this approach, algorithms can be trained to learn from input. Subsequent knowledge can be used to draw conclusions based on new information.

The basis of the system is deep learning techniques combined with distributed computing systems. This leverages both cloud and edge computing, allowing the technology to operate both on remote servers and close to where the data is generated (“the edge”).

Therefore, the computation is spread across multiple nodes closer to the data source to reduce cloud computing processes. This improves server response time and increases data security. Once collected, the data is transformed to estimate the angles that define the direction of the signal, providing a real-time update

The goal is to create a versatile system that uses the different technologies available and can be adapted to different potential use cases.

According to lead researcher Xavier Vilajosana: “There are numerous technological approaches today that try to exploit the characteristics of radio signals as a tool to obtain the relative position between objects.”

Specifically, Vilajosana adds: “This technological variety and the wide range of situations in which it can be used, with very diverse budgets and application environments, means that we need to develop a powerful framework for managing location data from different technologies in real time, which at the same time is capable of adapting to multiple industrial needs and is economically attractive”.


Please enter your comment!
Please enter your name here