Jesse Haviland

Robotics and AI Researcher

My research seeks to empower robots with the ability to operate reliably and adaptably in real-world settings. These environments extend far beyond the confines of controlled laboratory conditions, presenting dynamic, unstructured, and unpredictable challenges. To achieve this goal, my research adopts a highly interdisciplinary approach, bridging the fields of computer science, control systems, computer vision, and artificial intelligence. By exploring the confluence of these disciplines, I strive to create robots that can thrive in the complexity and uncertainty of the real world.

My research focus to date has been on developing novel control strategies for mobile manipulators (and table top manipulators) to move away from the typical stop-start, segmented (move base, then move arm paradigm) and clunky motion commonly shown by mobile manipulators. I have created algorithms that allow mobile manipulators to move in a coordinated whole-body human-like manner while also reacting to environment and goal changes in real-time.

about me

I am currently a Lecturer and Research Fellow at the Queensland University of Technology Centre for Robotics (QCR) and a Visiting Scientist at CSIRO Data 61 Robotics and Autonomous Systems Group.

My PhD, titled Control Strategies for Reactive Manipulation, was completed in 2022. This was undertaken at the Australian Centre for Robotic Vision and QCR under the supervision of Peter Corke, Niko Sünderhauf, and Feras Dayoub. This thesis was awarded the 2023 QUT Outstanding Doctoral Thesis Award and the 2023 Siganto Foundation Medal.

open source code

I write a lot of Python code for robotics. With Peter Corke, I started Python Robotics, an open-source robotics ecosystem made for Python. This started out in 2020 as a port of the famous Robotics Toolbox for MATLAB but has since grown into a much larger project. Collectively, packages within Python Robotics have over 750 000 downloads.

Some highlights of Python Robotics include:

• Robotics Toolbox for Python - Provides robot-specific functionality and contributes tools for representing the kinematics and dynamics of manipulators, robot models, mobile robots, path planning algorithms, kinodynamic planning, localisation, map building and simultaneous localisation and mapping.
• Spatial Math Toolbox for Python - Spatial Maths provides the ability to describe objects' position, orientation or pose in 2D or 3D spaces -- this capability underpins all of robotics. This package supports the Special Euclidean groups, Special Orthogonal groups, quaternions, twists, and various mathematical operations and conversions between them. Now being developed in collaboration with the Boston Dynamics AI Institute
• Swift - Swift is a robot simulator controlled by Python but displayed in a web browser using several web technologies, including React, Next.js, and Three.js. The entire simulator can be downloaded and installed from Python package managers PyPI or Conda-forge with no external downloads or installs.

teaching and education

In 2023, I created, coordinated and Lectured the subject Foundations of Kinematics and Algorithms in Robotics at QUT. This subject is a part of the new Robotics and AI masters course at QUT. I have released an open-source tutorial series on Spatial Mathematics to accompany this subject. I am also a lecturer for the subject Introduction to Robotics at QUT.

With Peter Corke, I created a two-part tutorial on manipulator differential kinematics (see Part I and Part II). The two tutorial articles are published in the IEEE Robotics and Automation Magazine. The entirety of the articles are accompanied by Python code through Jupyter Notebooks.