Beomjoon Kim

PhD Student
[Curriculum Vitae]
Learning and Intelligent Systems Group
{firstname} at


My goal is to enable robots to efficiently make decisions in complex environments by utilizing their past experience. In particular, my recent research focus on developing machine learning algorithms for robot task and motion planning problems, in which problems involve reasoning about both discrete, logical structures and continuous, geometric structures of the world. To this end, I use and develop tools such as generative models, deep learning, bandit algorithms, and reinforcement learning algorithms that are suitable for such problems. I am advised by Professors Tomas Lozano-Perez and Leslie Pack Kaelbling and I belong to Learning and Intelligent Systems Group at Massachusetts Institute of Technology

Peer-reviewed publications

  • Beomjoon Kim, Leslie Pack Kaelbling, Tomas Lozano-Perez. Guiding search in continuous state-action spaces by learning an action sampler from off-target search experience, AAAI Conference on Artificial Intelligence (AAAI), 2018. [paper]
  • Beomjoon Kim, Leslie Pack Kaelbling, Tomas Lozano-Perez. Learning to guide task and motion planning using score-space representation, International Conference on Robotics and Automation (ICRA), Winner of Best Cognitive Robotics Paper Award, 2017. [paper]
  • Beomjoon Kim, Leslie Pack Kaelbling, Tomas Lozano-Perez. Generalizing over uncertain dynamics for on-line trajectory generation, International Symposium of Robotics Research (ISRR) , 2015. [paper]
  • Beomjoon Kim, Joelle Pineau. Socially adaptive path planning in human environments using inverse reinforcement learning, International Journal of Social Robotics (SORO) , 2015. [paper]
  • Beomjoon Kim, Amir-massoud Farahmand, Doina Precup, Joelle Pineau. Learning from limited demonstrations , Neural Information Processing Systems (NIPS) , 2013. [paper]
  • Beomjoon Kim, Joelle Pineau. Maximum mean discrepancy imitation learning , Robotics: Science and Systems (RSS) , 2013. [paper]