Toward Human-Centred Safe Reinforcement Learning in the Real World

Project Member(s): Chen, L.

Funding or Partner Organisation: Australian Research Council (ARC Discovery Projects)
Australian Research Council (ARC Discovery Projects)

Start year: 2024

Summary: This project seeks to explore human-centred safe Reinforcement Learning (RL). Safe RL is a crucial field that enables the practical application of RL systems by fulfilling safety requirements. At present, safe RL assumes the presence of certain safety constraints in mathematical or logical forms. This project examines learning safety objectives from human-supplied data directly or language models indirectly, and implementing human-centred continuous corrections for safety enhancement. The resulting theories and algorithms will not only advance the frontiers of AI technology, but also contribute to the widespread application of safe RL, such as robotics and autonomous driving, bringing significant social and economic advantages.

FOR Codes: Artificial intelligence, Application software packages, Deep learning, Data mining and knowledge discovery

Acknowledgement of Country

UTS acknowledges the Gadigal people of the Eora Nation, the Boorooberongal people of the Dharug Nation, the Bidiagal people and the Gamaygal people, upon whose ancestral lands our university stands. We would also like to pay respect to the Elders both past and present, acknowledging them as the traditional custodians of knowledge for these lands.