Arnu Pretorius
Papers
1
Total Citations
12
H-Index
1
About
Arnu Pretorius is a leading researcher in multi-agent reinforcement learning (MARL) and robotics, with a particular focus on scaling complex coordination tasks. His most-cited work, "Scaling multi-agent reinforcement learning to full 11 versus 11 simulated robotic football" (2023, 12 citations), tackles the grand challenge of training autonomous agents to play realistic, full-team football—a problem long considered a benchmark for AI. This contribution demonstrates how MARL can move beyond toy environments to handle high-dimensional, real-world-inspired scenarios, where agents must learn cooperative strategies, spatial awareness, and dynamic role-switching. Pretorius’s research bridges the gap between theoretical reinforcement learning and practical deployment, often emphasizing sample efficiency and robust policy transfer. His work has implications for swarm robotics, autonomous driving, and any domain requiring decentralized decision-making. By pushing the boundaries of what MARL can achieve in competitive, multi-agent settings, Pretorius is helping to define the next generation of intelligent, collaborative systems.
Research Focus
Key Achievements
Top Papers
- 1