Jack Collins

Hi! Jack here. I am a postdoctoral researcher at the Oxford Robotics Institute within the Applied Artificial Intelligence Lab. My research is at the intersection of machine learning and robotics, in particular, looking at modelling the world for agents, task planning for long horizon tasks and machine introspection. I look to apply my work to real-world robots with the intention of enabling robots to interact and manipulate their environment, much like people can.

I completed my PhD at the Queensland University of Technology (QUT) whilst closely collaborating with the Commonwealth Scientific and Industrial Research Organisation (CSIRO). My research focused on the simulation of robots and how we could traverse the ‘Reality Gap’ using sim-to-real methods. The broad applicability of this work allows me to draw upon it today to transfer policies trained on simulated data to real-world platforms.

I also have a strong interest in Evolutionary Robotics. Instead of manually designing robots and controllers, Evolutionary Robotics investigates how we can apply evolutionary biology to evolve both a robot’s morphology and control.

What I Do

Robotics Research

Robotics Research

Conducting fundamental and applied research into the field of robotics with an emphasis on manipulation and an interest in bimanual and mobile manipulation. Experience in all aspects of research including academic writing, grant application, supervision and management.

ML Research

ML Research

Understanding of machine learning fundamentals and recent advances of state of the art. Applied knowledge in geometric deep learning including graph neural networks and generative models with an interest in task planning and world models.

Tutoring

Tutoring

Experience lecturing and tutoring subjects taught as part of engineering, robotics and computer science bachelor or masters degrees. Subjects that have been taught are found throughout the continuum from first year subjects to final year specialisations.

Resume

Experience

2021 - Current
University of Oxford

Postdoctoral Researcher

  • Postdoctoral researcher at the Oxford Robotics Institute in the Applied AI lab (A2I).
  • Research activities in areas including generative modelling, world models, manipulation (mobile, bimanual, dexterous), learning from demonstration, task planning, sim2real, deployment of robots to care homes and more.
  • Supervision of DPhil students and other lab responsibilities.
2024
King Abdullah University of Science and Technology (KAUST) Academy

Lecturer

Lecturer of KAUST Academy Advanced AI Course covering the foundations of convolutional neural networks and their application to computer vision, including image classification, segmentation, and object detection.

2022 – 2023
University of Oxford

Tutor

  • Tutor for C19 Machine Learning, a 4th year subject within the Masters in Engineering Science degree.
2022 – Current
Oxford College – Lady Margaret Hall

Programme Tutor

  • Lecturer and tutor of introductory and advanced machine learning courses. Advanced courses included programs on reinforcement learning, computer vision, unsupervised learning and graph neural networks. Tutoring including running seminars and tutorials on machine learning concepts and marking assessable items.
2019 – 2020
Holovision 3D

R&D Engineer

  • Lead developer of cobot welding cell. Deployed in 4 months for steel fabricator, Watkins Steel.
  • Exploration of the use of Mixed Reality in the construction industry through visualisation and annotation within Microsoft Hololens 1 and 2, Oculus Quest and Rift headsets.
  • Application of additive manufacturing methods to the construction industry.
2018 – 2021
Queensland University of Technology

Sessional Academic

  • Supervisor and Co-supervisor of undergraduate thesis students completing research in the robotics and manipulation domain.#
  • Tutor for a range of Electrical Engineering and Computer Science courses including:
    • Mechatronics Design 2 – Research, Design, and Implementation of an advanced mechatronic product.
    • Mechatronics Design 1 – Culmination of a product requiring mechanical design, electrical design, and embedded software.
    • Building IT Systems – Foundation unit taught in Python with topics including APIs, Databases, search patterns and GUIs.

Education

2018 - 2022
Queensland University of Technology / Commonwealth Scientific and Industrial Research Organisation

PhD Candidate

  • Thesis: Simulation to reality and back: A robot’s guide to crossing the reality gap
  • Supervisors: Dr. Juxi Leitner, Dr. David Howard and Dr. Ross Brown
2014 - 2017
Queensland University of Technology

Bachelor of Engineering(Mechatronics)(Hon)

  • Major in Robotics
  • 1st Class Honours
  • GPA – 6.25/7.0
  • Thesis Title: Automated Artificial Evolution of 3D Hexapod Legs

Robotic Technologies

ROS and associated packages

80%

Simulation (Rigid Body, Softy Body, Differentiable)

90%

Motion Capture

90%

Software Skills

Python

95%

C++

70%

Matlab

60%

Latex

95%

Git

90%

Software Containers (Singularity and Docker)

90%

High Performance Clusters (HPC)

90%

Publications

A Review of Differentiable Simulators

Rhys Newbury, Jack Collins, Kerry He, Jiahe Pan, Ingmar Posner, David Howard and Akansel Cosgun

[Paper] [Website]


D-Cubed: Latent Diffusion Trajectory Optimisation for Dexterous Deformable Manipulation

Jun Yamada, Shaohong Zhong, Jack Collins and Ingmar Posner

[Paper] [Website]


DreamUp3D: Object-Centric Generative Models for Single-View 3D Scene Understanding and Real-to-Sim Transfer

Yizhe Wu, Haitz Sáez de Ocáriz Borde, Jack Collins, Oiwi Parker Jones and Ingmar Posner

[Paper]


TWIST: Teacher-Student World Model Distillation for Efficient Sim-to-Real Transfer

TWIST: Teacher-Student World Model Distillation for Efficient Sim-to-Real Transfer

Jun Yamada, Marc Rigter, Jack Collins and Ingmar Posner

[Paper] [Website]


RAMP: A Benchmark for Evaluating Robotic Assembly Manipulation and Planning

Jack Collins*, Mark Robson*, Jun Yamada*, Mohan Sridharan, Karol Janik and Ingmar Posner

[Paper] [Website]


Leveraging Scene Embeddings for Gradient-Based Motion Planning in Latent Space

Jun Yamada*, Chia-Man Hung*, Jack Collins, Ioannis Havoutis and Ingmar Posner

[Paper]


Efficient Skill Acquisition for Complex Manipulation Tasks in Obstructed Environments

Jun Yamada, Jack Collins and Ingmar Posner

[Paper]


Taking Shape: A Perspective on the Future of Embodied Cognition and a new Generation of Evolutionary Robotics

David Howard, Jack Collins and Nicole Robinson

[Paper]


A Review of Physics Simulators for Robotic Applications

Jack Collins, Shelvin Chand, Anthony Vanderkop and David Howard

[Paper] [PDF]


Traversing the Reality Gap via Simulator Tuning

Jack Collins, Ross Brown, Jürgen Leitner and David Howard

[PDF]


Benchmarking Simulated Robotic Manipulation Through a Real World Dataset

Jack Collins, Jessie McVicar, David Wedlock, Ross Brown, David Howard and Jürgen Leitner

[Paper] [PDF] [Slides]


Comparing Direct and Indirect Representations for Environment-Specific Robot Component Design

Jack Collins, Ben Cottier and David Howard

[Paper] [PDF] [Slides]


Quantifying the reality gap in robotic manipulation tasks

Jack Collins, David Howard and Jürgen Leitner

[Paper] [PDF]


Towards the targeted environment-specific evolution of robot components

Jack Collins, Wade Geles, David Howard and Frederic Maire

[Paper] [PDF] [Slides]


Contact

Get in Touch

Oxford, United Kingdom

jack@jacktcollins.com

Consultations Available

How Can I Help You?