Application of radar sensing in Kovilta Oy sensor framework
Paasio, Kalle (2024)
Paasio, Kalle
2024
Sähkötekniikan DI-ohjelma - Master's Programme in Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2024-03-06
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202402062141
https://urn.fi/URN:NBN:fi:tuni-202402062141
Tiivistelmä
This work explores radar sensing for sensor fusion applications with a camera. It addresses the fundamentals of radar, sensor fusion, and filtering. These fundamentals are put into practice in the implementation phase, where a physical system is constructed. The purpose of this work is to explore for Kovilta Oy a previously uncharted territory of radar sensing. The demonstration platform provides a real-world view into data formats, algorithms, radio-frequency related phenomena, and communication interfaces. It also serves as a basis for further in-house development being the first, and thus a foundational element, of Kovilta Oy’s radar environment.
The implemented system is capable of tracking multiple humans in a room while continuously calculating their speed and position vectors. When applied to vehicular tracking, it can distinguish people, cars, cyclists and stop signs. In the conducted tests, the system was able to track cars from a distance of up to 50 meters. The novel sensor fusion algorithm also has the ability to scale its reliability with better radar platforms and with increasing computation power. Systems like this can be used as guidance for simultaneous localization and mapping, advanced driver assistance, or as standalone object-level monitoring systems. In particular, Kovilta Oy is interested in applying these systems to autonomous vehicles such as agricultural drones and self-driving cars.
In the final chapters, avenues for further exploration are presented including alternative approaches for sensor fusion, new radar modulation schemes, and ways to utilize machine learning on radar data. The proceedings of this work are made Open Source and are available on the author’s GitHub under the MIT license.
The implemented system is capable of tracking multiple humans in a room while continuously calculating their speed and position vectors. When applied to vehicular tracking, it can distinguish people, cars, cyclists and stop signs. In the conducted tests, the system was able to track cars from a distance of up to 50 meters. The novel sensor fusion algorithm also has the ability to scale its reliability with better radar platforms and with increasing computation power. Systems like this can be used as guidance for simultaneous localization and mapping, advanced driver assistance, or as standalone object-level monitoring systems. In particular, Kovilta Oy is interested in applying these systems to autonomous vehicles such as agricultural drones and self-driving cars.
In the final chapters, avenues for further exploration are presented including alternative approaches for sensor fusion, new radar modulation schemes, and ways to utilize machine learning on radar data. The proceedings of this work are made Open Source and are available on the author’s GitHub under the MIT license.