Arbitrary motion Synthetic Aperture Radar

Abstract (Swedish)

Syftet med denna avhandling är att utveckla en ny metod för att producera bilder med syntetisk aperturradar (SAR), med utgångspunkt i scenarier med arbiträr rörelse vad gäller radarsensorn. SAR är en väletablerad metod för att skapa 2- eller 3-dimensionella radarbilder, som traditionellt sett antar att radar-sensorns rörelse är linjär och förutsägbar. Därmed är positionerna för varje radardata-punkt kända och följer ett fördefinierat mönster. Denna avhandling utforskar emellertid nya metoder som gör det möjligt för radarn att röra sig fritt.

Vi har utvecklat en prototyp baserad på Pulsed Coherent Radar-sensorn A121 från Acconeer för att demonstrera potentialen och prestandan i vår metod. Det resulterande systemet kan framgångsrikt skapa en bild av sin omgivning och upptäcka hinder i 3D-rymden. Med hjälp av denna prototypuppställning har vi utvärderat olika konfigurationer och optimeringar av algoritmen, samt konfigurationer för radarsensorn, inklusive val av lins. Vi har testat uppställningen för många olika potentiella objekt och utvärderat dess prestanda i flera olika miljöer.

Systemet förlitar sig på en ström av exakta poseringsdata för radar-sensorn. Två olika system utforskades för detta ändamål. Först en Inertial Measurement Unit (IMU), och sedan VR-hårdvara som bygger på IR-fyr-spårning. Tidiga experiment visade mindre lovande resultat för IMU-baserad spårning, och denna väg ansågs inte vara värd att utforska vidare. Att spåra sensorn med hjälp av en VR-uppsättning visade emellertid lovande resultat, och fler experiment genomfördes för att utvärdera användbarheten och begränsningarna i dess användning för att skapa radarbilder. Dessa två tillvägagångssätt är dock en delmängd av ett stort antal potentiella spårningssystem.

Våra resultat visar att SAR är möjligt att implementera i fall där radarns rörelse är arbiträr, förutsatt att sensorns position kan bestämmas med viss noggrannhet för alla insamlade radardatapunkter. Det visar också hur SAR baserad på arbiträr rörelse potentiellt kan möjliggöra fler användningsområden för radar-sensorer, till exempel inom VR och robotik för kartläggning av miljöer och detektion av objekt.

Abstract

The purpose of this thesis is to develop a novel approach to producing Synthetic Aperture Radar (SAR) images, assuming the scenario of arbitrary motion in regards to the radar sensor. SAR is a well-researched method for creating 2- or 3-dimensional radar images, traditionally assuming the radar sensor's motion to be linear and highly predictable. Thus the locations where each radar data is sampled are known and follow a predefined pattern. This thesis, however, explores new methods that enable the radar to move freely.

We have developed a prototype based on the A121 Pulsed Coherent Radar sensor from Acconeer to demonstrate the performance and use-case potential of the method we came up with. The resulting system can successfully create an accurate map of its surroundings, detecting obstacles in 3D space. Using this prototype setup, we have evaluated different configurations and kinds of optimizations for the algorithm developed, as well as configurations for the radar sensor, including lens choice. We have tested the setup for many different potential targets and evaluated its performance in several different environments.

The system relies on a stream of accurate pose data in regard to the radar sensor. Two different systems were explored for this purpose. Firstly, the use of an Inertial Measurement Unit (IMU), and secondly, using the IR lighthouse tracking from virtual reality hardware. Early experiments showed less than promising results for IMU-based tracking, and this path was not deemed worth exploring further. Tracking the sensor using a VR setup, however, showed promising results and more experiments were performed to evaluate the viability and limitations of its use in creating radar images. These two approaches, though, are a subset of a vast number of potential pose-tracking systems.

The final results show that SAR is possible to implement in cases where the motion of the radar is arbitrary, given the position of the sensor can be determined to some degree of accuracy for all collected radar data points. It also demonstrates how arbitrary motion SAR could potentially enable more use cases for radar sensors, for example, in VR, robotics for the mapping of environments, and object detection.

Popular Abstract

Radar is a technology that uses radio waves to detect and locate objects in its vicinity. In its most simple form, it works by emitting a radio wave signal and measuring the time it takes for the signal to bounce back after it has hit an object, much like how sonar works with sound. This information is then used to determine the distance, speed, and other characteristics of the object. More sophisticated radar technologies, though, have a wide range of use cases, from military applications to weather forecasting and air traffic control. In military applications, radar is used for surveillance, target acquisition, and weapon guidance. Weather radar is used to track and forecast storms, while air traffic control radar tracks the location and speed of aircraft in flight.

Acconeer AB, a Malmö-based company started in 2012 around research into Pulsed Coherent Radar (PCR) technology at LTH, develops a now commercially available 60GHz radar sensor with a small footprint, approachable price, and small power consumption. Its A111 and A121 radar sensors are seeing increasing adoption by a plethora of tech giants and independent inventors alike for a multitude of use cases. Recently, Volvo announced Acconeer radar-based interior presence detection will be featured in some upcoming car models. Consumer electronic use cases like this are a very novel concept, just springing up in the last decade, and there is yet a lot to explore in this field.

In recent years, there have been several modern developments in radar technology. One of these is the use of Synthetic Aperture Radar (SAR) which, for example, allows for the creation of high-resolution images of the Earth's surface by using a radar system on a moving platform, such as a satellite or airplane. SAR is a type of radar that uses the motion of the radar antenna to create images containing a lot of information that would not be possible to achieve with a single radar sweep. The antenna motion allows the radar to gather data from different positions and angles, which are combined to create a synthetic aperture. The concept has also been shrunken to the Acconeer PCR scale to produce images of objects at a more human scale. There is ongoing research into using radar for applications such as autonomous driving, where radar sensors can be used to detect and avoid obstacles on the road.

In the VR-space there is a constant push to enhance the user experience through immersion, which has pushed technology far in terms of different performance metrics like latency, resolution, and precision. There is, however, still the risk of breaking immersion by, for example, bumping into an obstacle in the environment. This has prompted research into technology to blur the lines between VR and AR by making VR users aware of their real-world surroundings. This has been tackled in different ways, for example, by using RGB-D cameras and LiDAR.

In this study, we have explored novel ways of using radar for SAR in cases where the motion of the radar sensor is unpredictable. We have implemented this successfully in the specific use case of obstacle detection in the VR application.

• author: Alberius, Filip and Rolander, Erik
• supervisor: Andreas Jakobsson
• organization: Mathematical Statistics
• course: FMSM01 20231
• year: 2023
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: RADAR, SAR, Synthetic Aperture Radar, Pulsed Coherent Radar, PCR, VR, Virtual Reality, IMU, Inertial Measurment Unit, VAA, Virtual Antenna Array, Obstacle detection, Signal Processing, Matched Filter Method.
• publication/series: Master’s Theses in Mathematical Sciences
• report number: LUTFMS-3470-2023
• ISSN: 1404-6342
• other publication id: 2023:E13
• language: English
• id: 9114031
• date added to LUP: 2023-05-05 13:49:09
• date last changed: 2023-05-11 14:38:25