Jess Jones

University of Bristol
FARSCOPE CDT
jess.jones@bristol.ac.uk

PhD candidate in Robotics and Autonomous Systems at the University of Bristol (FARSCOPE CDT), with a First Class BSc in Computer Science from University of Bristol.

I am interested in deep learning, multi-agent reinforcement learning, and integrating AI as a tool for robotics, and to support the scientific process.

I continue to work as a developer and researcher at CASM Technology on the analysis of online content with specific focus on misinformation through the improper use of Generative AI material.

As an active a member of the PROTEAS group, I enjoy exploring new opportunities to develop autonomous systems for positive environmental and social impact. I am currently involved in building software to support the education of young children with respect to local environmental habitats.

Before moving into tech and research full time, I held leadership roles in hospitality and a vertical-farming start-up, which shaped how I approach teamwork and delivery.

Supervisors: Prof. Sabine Hauert, OBE and Prof. Raul Santos-Rodriguez, University of Bristol.

news

Jun 05, 2026	Paper accepted at ICPR: CVBMC 2026 Workshop: CoralLite: μCT Reconstruction of Coral Colonies from Individual Corallites (arXiv preprint).
Dec 19, 2025	Paper accepted at AAMAS 2026: Assessing VLM-Driven Semantic Affordance Inference for Non-Humanoid Robot Morphologies (arXiv preprint, DOI).
Sep 01, 2023	Started PhD in Robotics & Autonomous Systems at the University of Bristol, supervised by Prof. Sabine Hauert, OBE and Prof. Raul Santos-Rodriguez through the FARSCOPE CDT.
Aug 01, 2023	Awarded the Hargrave Prize for best final-year BSc thesis and the NAG HPC Student Prize at the University of Bristol.

selected publications

AAMAS 2026

Assessing VLM-Driven Semantic Affordance Inference for Non-Humanoid Robot Morphologies

Jess Jones, Raul Santos-Rodriguez, and Sabine Hauert

In Proceedings of the International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2026

Abs DOI arXiv HTML

Investigates whether VLMs can infer affordances for non-humanoid robot morphologies using a hybrid real-world and synthetic dataset, finding promising generalisation but inconsistent performance across object domains and a conservative bias with high false negative rates.

ICPR CVBMC 2026

CoralLite: μCT Reconstruction of Coral Colonies from Individual Corallites

Jess Jones, Leonardo Bertini, Kenneth Johnson, and 2 more authors

In Proceedings of the CVBMC Workshop at the International Conference on Pattern Recognition (ICPR), 2026

Abs arXiv

Proposes CoralLite, a first deep learning baseline for reconstructing individual corallites in 3D, combining volumetric segmentations with cross-slice linking from single slices to colony scale.