Training humanoid robots for high-risk applications using large and small AI models By Michelle LucaJune 17th, 2026 Project overview Project title Training humanoid robots for high-risk applications using large and small AI models Project description This Project aims to develop trustworthy AI systems for humanoid robots operating in hazardous environments such as explosive ordnance disposal and high-risk mining and construction tasks. The project investigates how computer vision, large language models, reasoning methods, and reinforcement learning can be integrated to enable robots to interpret high-level mission goals, plan safe action sequences, and execute dexterous manipulation tasks. Computer vision techniques will play a key role in enabling scene understanding, object detection, and spatial reasoning from multimodal sensor data, forming the foundation for robust perception in real-world settings. Supervisory team University Name of university supervisorZhe HouName of universityGriffith UniversityEmail addressz.hou@griffith.edu.auFacultySchool of ICT CSIRO Name of CSIRO supervisorQin WangEmail addressqin.wang@csiro.auCSIRO Research UnitTechnology Industry Name of industry supervisorStephen BornsteinName of organisationCyborg Dynamics Engineering Pty LtdEmail addresss.bornstein@cyborgdynamicseng.com Further details Primary location of studentGriffith University, 170 Kessels Road, Nathan QLD 4111, Australia Industry engagement component locationCyborg Dynamics Engineering Pty Ltd, Unit 4, 20 Smallwood Place, Murarrie QLD 4172, Australia Other locationsCSIRO Marsfield, 26 Pembroke Road, Marsfield NSW 2122, Australia Ideal student skillsetEssential: Strong background in artificial intelligence, computer science and robotics, with experience in machine learning or reinforcement learning. Proficiency in Python and modern ML frameworks.Demonstrated research ability.Strong problem-solving skills.Ability to collaborate with industry engineers and communicate technical results clearly. Desirable: Knowledge of control systems.Application close dateOpen until position filledApplyContact Zhe Hou Related Previous post:Assessing technology innovations in Australian temperate region beef and sheep systems Coming up next:End of list