In Defence of System 2: Human-Inspired Cognitive Intelligence for Reasoning and Robotics
Speaker:
A/Prof. Hamid Rezatofighi, Monash University, Australia
Online Teams meeting:
Monday, 30 Nov 2026 at 11.30am
Host: IEEE Victorian Section SMC Chapter and Swinburne University of Technology. Organized by A/Prof. Hailing Zhou and Prof. Saeid Nahavandi
Abstract:
Recent advances in neural foundation models have transformed artificial intelligence, enabling remarkable capabilities in visual perception, language understanding, and robotic control. Yet, despite the growing emphasis on scaling models, data, and computation, fundamental challenges remain in complex reasoning, compositional generalisation, and interpretable decision-making. Inspired by human cognition, this raises an important question: Can fast, intuitive, System 1-like neural foundation models alone deliver the sophisticated intelligence required for autonomous systems operating in complex, dynamic environments?
In this talk, I will advocate for a complementary direction inspired by System 2 cognition: slower, deliberate, structured, and compositional reasoning. I will present our recent research at Monash on agentic, compositional neuro-symbolic AI, combining the strengths of neural foundation models in perception and language understanding with explicit world representations, logical reasoning, memory, and planning. Rather than relying exclusively on monolithic end-to-end neural architectures, our approach explores how specialised intelligent agents and structured reasoning mechanisms can work together to address complex visual reasoning and decision-making challenges.
I will highlight some of our recent research contributions that expose limitations of current neural foundation models and demonstrate how agentic neuro-symbolic approaches can advance visual understanding, compositional reasoning, and adaptive decision-making. I will also showcase how these principles extend to robotic autonomy, drawing on our DARPA- and ONR-supported research, including autonomous robotic systems capable of complex missions under uncertainty and dynamic environmental conditions.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
- John St, Hawthorn VIC
- Melbourne, Victoria
- Australia 3122
- Building: EN
- Room Number: 609A
- Contact Event Host
- Co-sponsored by Swinburne University of Technology
- Starts 09 November 2026 05:51 PM UTC
- Ends 29 November 2026 01:00 PM UTC
- No Admission Charge
Speakers
Hamid of Monash University
In Defence of System 2: Human-Inspired Cognitive Intelligence for Reasoning and Robotics
Biography:
Associate Professor Hamid Rezatofighi is an AI and robotics researcher at Monash University, Australia, where he joined as a Lecturer in 2020. His research spans computer vision, machine learning, robotics, and neuro-symbolic AI, with a focus on visual perception and reasoning, embodied intelligence, and Physical AI.
Prior to joining Monash, he was an Endeavour Research Fellow at Stanford University and a Senior Research Fellow at the Australian Institute for Machine Learning (AIML), University of Adelaide. He received his PhD from the Australian National University in 2015.
He has authored approximately 120 publications in leading international venues, attracting over 22,000 citations. His research has attracted more than $18 million in competitive funding, including major DARPA- and ONR-supported projects. His achievements have been recognised through the ONR Global-X Challenge Award, the Australian Government Endeavour Research Fellowship, and the Monash Faculty of IT Dean's Award for Early Career Research Excellence.
He actively serves the international research community as Area Chair or Lead Area Chair for major conferences, including CVPR, NeurIPS, ICML, ICLR, ECCV, ICCV, AAAI, and IJCAI. His editorial appointments include Senior Associate Editor of IEEE Transactions on Image Processing, Editor for IROS, and Associate Editor for ICRA and the Artificial Intelligence Journal.
Email: