Anssi Kanervisto

Finland University, University of Eastern Finland

Papers

Total Citations

H-Index

About

Anssi Kanervisto is a researcher at the forefront of imitation learning and human-AI interaction, with a focus on bridging the gap between simulated environments and real-world robotics. His work centers on teaching machines to replicate complex human behaviors, particularly in video games and robotic control tasks. Kanervisto’s most notable contribution is his pioneering study on using diffusion models for imitating human behavior in sequential environments, a paper that has already garnered 23 citations since its 2023 publication. This work addresses the challenge of human behavior’s stochastic and multimodal nature, offering a novel approach to observation-to-action modeling. He has also made significant strides in benchmarking end-to-end behavioural cloning on video games, providing foundational insights into how computers can learn from human demonstrations without reinforcement learning. Additionally, Kanervisto has explored the transfer of learned policies between video games and real robots, tackling the critical issue of action space differences. His research is instrumental in advancing practical applications of AI, from gaming to robotics, and his work continues to shape how we train autonomous systems to interact with the world.