Practical investigations in robot localization using ultra-wideband sensors

Wright, Nicholas Albert Wyatt

Permalink

https://hdl.handle.net/2142/110610 Copy

Description

Title

Practical investigations in robot localization using ultra-wideband sensors

Author(s)

Wright, Nicholas Albert Wyatt

Issue Date

2021-04-30

Director of Research (if dissertation) or Advisor (if thesis)

Dullerud, Geir E

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• robot localization
• ultra-wideband
• UWB
• sensors
• ROS
• Kalman filter
• pose estimate
• phase difference of arrival
• PDoA
• multilateration
• trilateration
• motion capture
• ground truth
• GNSS-denied
• GPS-denied
• total station

Abstract

Robot navigation is rudimentary compared to the capabilities of humans and animals to move about their environments. One of the core processes of navigation is localization, the problem of answering where one is at the present time. Robot localization is the science of using various sensors to inform a robot of where it is within its environment. Ultra-wideband (UWB) radio is one such sensor technology that can return absolute position information. The algorithm to accomplish this is known as multilateration, which uses a collection of distance measurements between multiple robot tag and environment anchor pairs to calculate the tag’s position. UWB is especially suited to the task of returning precise distance measurements due to its capabilities of short duration, high amplitude pulse generation and detection. Decawave Ltd. has created an UWB integrated circuit to perform ranging and a suite of products to support this technology. Claimed and verified accuracies using this implementation are on the order of 10cm. This thesis describes various experiments carried out using Decawave technology for robot localization. The progression of the chapters starts with commercial product verification before moving into development and testing in various environments of an open-source driver package for the Robot Operating System (ROS), then the development of a novel phase difference of arrival (PDoA) sensor for three-dimensional robot localization without an UWB anchor mesh, before concluding with future research directions and commercialization potential of UWB. This thesis is designed as a compilation of all that the author has learned through primary and secondary research over the past three years of investigation. The primary contributions are: 1. A modular ROS UWB driver framework and series of ROS bags for offline experimentation with multilateration algorithms. 2. A robust ROS framework for comparing motion capture system (MoCap) ground truth vs sensor data for rigorous statistical analysis and characterization of multiple sensors. 3. Development of a novel UWB PDoA sensor array and data model to allow 3D localization of a target from a single point without the deployment of an antenna mesh.

Graduation Semester

2021-05

Type of Resource

Thesis

Permalink

http://hdl.handle.net/2142/110610

Copyright and License Information

Copyright 2021 Nicholas AW Wright

Owning Collections