Papers
15
Total Citations
243
H-Index
9
About
David Budden is a robotics and machine learning researcher whose work spans data-driven robotics, reinforcement learning, and computer vision. His most significant contributions center on scalable frameworks for robot learning, particularly the development of batch reinforcement learning systems that leverage large datasets of recorded robot experience combined with learned reward functions — work that has collectively accumulated over 100 citations and demonstrated real-world applicability across multiple object manipulation tasks. His research on adversarial imitation learning further advanced the field by identifying and addressing a critical vulnerability: the tendency of discriminator networks to fixate on task-irrelevant visual features, undermining reward signal quality. Earlier in his career, Budden made foundational contributions to humanoid robot soccer through the RoboCup simulation leagues, developing novel approaches to ball detection, particle filtering for robot localisation, and unsupervised colour recognition for real-time image processing. These contributions helped establish replicable benchmarks for evaluating complex robotic systems. His trajectory reflects a natural evolution from applied robotics perception to large-scale reinforcement learning, making his body of work particularly valuable for researchers working at the intersection of robot autonomy, imitation learning, and scalable AI systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3A Novel Approach to Ball Detection for Humanoid Robot Soccer23 citations · 2012
- 4Task-Relevant Adversarial Imitation Learning22 citations · 2019
- 5
- 6
- 7
- 8A Framework for Data-Driven Robotics11 citations · 2019
- 9Unsupervised Recognition of Salient Colour for Real-Time Image Processing11 citations · 2014
- 10