scope for the student to shape the direction around the problems that most excite them. To ground this work, the project uses cooperative manipulation and transport as a testbed: a setting where several agents must jointly move or reshape an object that none can handle alone. Here the agents are physically coupled through the object they share, so coordination must happen through the forces and contact they exert — a problem of embodied intelligence, and a richer, more realistic challenge than abstract MARL benchmarks. Depending on the student's interests, directions might include new algorithms for credit assignment or scalability, decentralised coordination from local sensing, or the integration of learning with control-theoretic guarantees of stability and safety. Your work can be developed and tested quickly and at scale in simulation, with the opportunity to validate the most promising methods on physical robots. We welcome candidates with a background in AI, machine learning, mechatronics, control, robotics, mechanical engineering, electronic engineering, computer science, or a related field. You should have an interest in reinforcement learning and multi-agent systems, strong programming and mathematical skills, and enjoy designing and testing algorithms at scale in simulation. Experience with robots is beneficial but not required. Informal enquiries, with a CV and short statement of interest, are welcome to Dr Kai-Fung Chu (kaifung.chu@bristol.ac.uk).

scope for the student to shape the direction around the problems that most excite them. To ground this work, the project uses cooperative manipulation and transport as a testbed: a setting where several agents must jointly move or reshape an object that none can handle alone. Here the agents are physically coupled through the object they share, so coordination must happen through the forces and contact they exert — a problem of embodied intelligence, and a richer, more realistic challenge than abstract MARL benchmarks. Depending on the student's interests, directions might include new algorithms for credit assignment or scalability, decentralised coordination from local sensing, or the integration of learning with control-theoretic guarantees of stability and safety. Your work can be developed and tested quickly and at scale in simulation, with the opportunity to validate the most promising methods on physical robots. We welcome candidates with a background in AI, machine learning, mechatronics, control, robotics, mechanical engineering, electronic engineering, computer science, or a related field. You should have an interest in reinforcement learning and multi-agent systems, strong programming and mathematical skills, and enjoy designing and testing algorithms at scale in simulation. Experience with robots is beneficial but not required. Informal enquiries, with a CV and short statement of interest, are welcome to Dr Kai-Fung Chu (kaifung.chu@bristol.ac.uk).