Indoor positioning with AI/ML using simulated 5G data
(2024) EITM01 20241Department of Electrical and Information Technology
- Abstract
- With future technologies and Industry 4.0, the need for robust and accurate positioning to optimize productivity and industrial operations increases. The previous positioning methods using triangulation or angle-of-arrival are not good enough, especially not for clutter-dense factories with poor line-of-sight conditions.
For this thesis project, a convolutional neural network model was trained to better estimate time-of-arrival and classify line-of-sight for a clutter-sparse simulated factory. For the clutter-dense factory, a residual network fingerprinting model was trained to map channel impulse responses to a position. Both the clutter-sparse and clutter-dense factories were modified by moving and rotating machines as well as adding... (More) - With future technologies and Industry 4.0, the need for robust and accurate positioning to optimize productivity and industrial operations increases. The previous positioning methods using triangulation or angle-of-arrival are not good enough, especially not for clutter-dense factories with poor line-of-sight conditions.
For this thesis project, a convolutional neural network model was trained to better estimate time-of-arrival and classify line-of-sight for a clutter-sparse simulated factory. For the clutter-dense factory, a residual network fingerprinting model was trained to map channel impulse responses to a position. Both the clutter-sparse and clutter-dense factories were modified by moving and rotating machines as well as adding robots in order to investigate the robustness to changes in the factory environment.
The factories simulated production areas, assembly areas, beam structures and robots using the Ericsson state-of-the-art version of Nvidia Omniverse.
The fingerprinting model was also used in a real scenario from Mobile World Congress 2024, where a robot drove two routes in an open office environment.
The results show that both neural networks gave a positioning error below 1 m. For the modified scenarios, the convolutional neural network was more robust than the residual network fingerprinting model. The modifications mainly impacted positioning errors locally where changes were introduced. (Less) - Popular Abstract
- Industrial processes are moving toward automation, in what is called Industry 4.0, the fourth industrial revolution. For factories and warehouses, the need for efficiency and precision is paramount.
For instance, precise indoor positioning of autonomous vehicles or manufactured goods.
The use of Artificial Intelligence (AI) could be a way to facilitate improved positioning accuracy and adaptibility for these needs.
In factory environments, machinery and equipment are often densely deployed with other obstructions like pillars and beam structures. For such scenarios with limited visibility, conventional positioning solutions face challenges in providing accurate positioning. Consequently, leveraging AI methods to address issues of... (More) - Industrial processes are moving toward automation, in what is called Industry 4.0, the fourth industrial revolution. For factories and warehouses, the need for efficiency and precision is paramount.
For instance, precise indoor positioning of autonomous vehicles or manufactured goods.
The use of Artificial Intelligence (AI) could be a way to facilitate improved positioning accuracy and adaptibility for these needs.
In factory environments, machinery and equipment are often densely deployed with other obstructions like pillars and beam structures. For such scenarios with limited visibility, conventional positioning solutions face challenges in providing accurate positioning. Consequently, leveraging AI methods to address issues of conventional solutions is a growing interest.
This thesis has been conducted by creating two baseline factory environments, designed to have different line-of-sight conditions.
The simulated baseline factories have been modified by introducing changes to the physical arrangement of machines and equipment.
Simulated 5G data for the factory environments was used to train two AI models, learning patterns in the data to do positioning predictions.
Besides evaluating positioning for the two baselines, the models' capability to generalize to modified factory environments is investigated. The generalization aspect is of interest as it would provide robust and flexible solutions for autonomous industries, aligning with the Industry 4.0 transformation.
The positioning results are promising and lead to significant improvements compared to conventional solutions, especially for scenarios with poor line-of-sight conditions. Moreover, the generalization aspect is assessed to do quite well, depending on how drastic the changes are.
Towards the end, real 5G data measurements from an office environment is evaluated with one of the AI models. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9166486
- author
- Söderbom, Philip LU and Nilsson, Jonathan LU
- supervisor
- organization
- course
- EITM01 20241
- year
- 2024
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Indoor positioning, 5G, Industry 4.0, Simulations, AI, ML, Neural Networks
- report number
- LU/LTH-EIT 2024-1002
- language
- English
- id
- 9166486
- date added to LUP
- 2024-06-26 11:12:50
- date last changed
- 2024-06-26 11:12:50
@misc{9166486,
abstract = {{With future technologies and Industry 4.0, the need for robust and accurate positioning to optimize productivity and industrial operations increases. The previous positioning methods using triangulation or angle-of-arrival are not good enough, especially not for clutter-dense factories with poor line-of-sight conditions.
For this thesis project, a convolutional neural network model was trained to better estimate time-of-arrival and classify line-of-sight for a clutter-sparse simulated factory. For the clutter-dense factory, a residual network fingerprinting model was trained to map channel impulse responses to a position. Both the clutter-sparse and clutter-dense factories were modified by moving and rotating machines as well as adding robots in order to investigate the robustness to changes in the factory environment.
The factories simulated production areas, assembly areas, beam structures and robots using the Ericsson state-of-the-art version of Nvidia Omniverse.
The fingerprinting model was also used in a real scenario from Mobile World Congress 2024, where a robot drove two routes in an open office environment.
The results show that both neural networks gave a positioning error below 1 m. For the modified scenarios, the convolutional neural network was more robust than the residual network fingerprinting model. The modifications mainly impacted positioning errors locally where changes were introduced.}},
author = {{Söderbom, Philip and Nilsson, Jonathan}},
language = {{eng}},
note = {{Student Paper}},
title = {{Indoor positioning with AI/ML using simulated 5G data}},
year = {{2024}},
}