PhD Researcher in AI and Robotics
Doug Tilley
University of Bath | Human-Robot Collaboration | Founder at DotResearch
I design and evaluate wearable and assistive robotics systems, with a focus on practical human-robot interaction, robust control, and efficient machine learning for deployment.
- Current Role PhD Candidate in ART-AI, University of Bath
- Core Theme Multimodal Deep Learning for Human-Robot Interaction
- Industry and Applied Work TinyML, Robotics Simulation, Multimodal Learning, Optimization, Wildlife Conservation
Currently Building
From neuromorphic architectures to field-deployed wildlife sensors — here's what's on the bench.
Multimodal Human Activity Recognition
Hierarchical deep learning architectures integrating vision, biosignals, and motion capture for robust activity and postural transition recognition.
PublicationsNeuromorphic and Explainable Architectures
Hybrid SNN-SSM architectures for explainable temporal modeling, with novel representation techniques with biologically plausible optimizers.
Research OverviewLatent Analysis for Robotics actions and rewards
Mechanistic interpretability and Latent space representation methods applied to robotics domains, ongoing research at the National Institute of Informatics (NII), Tokyo.
NII CollaborationDotResearch — TinyML for Conservation
Co-founded startup deploying patent-pending embedded vision sensors for wildlife monitoring. Custom GenAI pipelines for domain adaptation, deployed across 3 countries in harsh environments.
Visit DotResearchSoft Wearable Exoskeletons for Rehabilitation
Hierarchical control of soft wearable exoskeletons to assist postural transitions in stroke patients, integrating multimodal sensing with adaptive actuation.
PublicationsCode & Projects
Active repositories spanning optimizers, spiking networks, embedded vision, and more. Some are pre-publication or in the works.
Loading repositories...
Selected Publications
Highlights from my publication record. Full list and citation metrics on Google Scholar.
Google Scholar Profile
Full publication list, citation metrics, and co-author network.
Hierarchical Deep Learning for Human Activity Recognition Integrating Postural Transitions
Hierarchical CNN-LSTM architecture for activity recognition with postural transition integration. Invited Best Paper of IEEE Sensors 2023.
BEATRIX: An Open Source Humanoid Head Platform
Open-source humanoid head platform for robotics teaching and HRI research.
Ionic Charge Emission Into Fog From a Remotely Piloted Aircraft
Experimental study of ion release effects on fog droplet properties using an instrumented UAV with corona charge emitters.
Research Interests
The threads that connect my work; from biologically plausible learning to robots that understand people.
Multimodal Deep Learning
Hierarchical and fusion architectures integrating vision, language, and biosignal modalities for human-robot interaction.
Neuromorphic and Explainable AI
Spiking neural networks, hybrid SSM architectures, and interpretability methods for temporal and noisy data.
Optimizers and Training Dynamics
Novel optimizers for noisy and non-stationary loss landscapes, combining trajectory-aware adaptation with sharpness-aware methods.
Efficient AI for Edge and Embedded Systems
TinyML deployment, model compression, and custom architectures for resource-constrained platforms.
Field Robotics and Rapid Prototyping
UAV systems, sensor fusion, and end-to-end deployment from concept to validated field prototype.
Vision-Language-Action Models
Sim-to-real reinforcement learning and VLA architectures for embodied AI and manipulation tasks.