Miroslav Bogdanovic

Miroslav Bogdanovic

I'm a Postdoctoral Fellow at the University of Toronto, Vector Institute, and Acceleration Consortium. Previously, I was at the Max Planck Institute for Intelligent Systems and the University of Oxford.

Email / Scholar

Publications My research interests include precise and reactive robotic manipulation, learning from demonstration (BC) and reinforcement learning (RL), large-scale data generation in simulation and sim-to-real transfer, and the use of foundation models in robotics.
	AnyPlace: Learning Generalized Object Placement for Robot Manipulation Yuchi Zhao, Miroslav Bogdanovic, Chengyuan Luo, Steven Tohme, Kourosh Darvish, Alán Aspuru-Guzik, Florian Shkurti, Animesh Garg CoRL, 2025 project page / PDF / arXiv / code A two-stage approach for object placement, utilizing a vision-language model to identify local placement region, allowing pose-prediction module trained on purely synthetic, local-only data to generalize to the real world.
	ORGANA: A Robotic Assistant for Automated Chemistry Experimentation and Characterization Kourosh Darvish, Marta Skreta, Yuchi Zhao, Naruki Yoshikawa, Sagnik Som, Miroslav Bogdanovic, Yang Cao, Han Hao, Haoping Xu, Alán Aspuru-Guzik, Animesh Garg, Florian Shkurti Matter, 2025 project page / PDF / arXiv / video An integrated system for chemistry lab automation that designs the experiment protocol through interaction with the user, then executes it using a fully automated robotics pipeline.
	Automated Planning Domain Inference for Task and Motion Planning Jinbang Huang, Allen Tao, Rozilyn Marco, Miroslav Bogdanovic, Jonathan Kelly, Florian Shkurti ICRA, 2025 PDF / arXiv A method for automatic inference of planning domains, allowing us to go from only a few demonstrations of a human doing the task to a robot solving new instances autonomously.
	CLIMB: Language-Guided Continual Learning for Task Planning with Iterative Model Building Walker Byrnes, Miroslav Bogdanovic, Avi Balakirsky, Stephen Balakirsky, Animesh Garg ICRA, 2025 project page / PDF / arXiv A continual learning approach for task planning that solves a curriculum of tasks of increasing complexity, expanding and improving an internal logical world model as it does so.
	Model-free Reinforcement Learning for Robust Locomotion using Demonstrations from Trajectory Optimization Miroslav Bogdanovic, Majid Khadiv, Ludovic Righetti Frontiers in Robotics and AI, 2022 project page / PDF / arXiv / video A reinforcement-learning-based pipeline in simulation, allowing us to go from a single generated trajectory to robust dynamic behavior that can be directly deployed on a real robot.
	Learning Variable Impedance Control for Contact Sensitive Tasks Miroslav Bogdanovic, Majid Khadiv, Ludovic Righetti RA-L, 2020 PDF / arXiv / video Learning adaptive impedance control policies and showing how they improve robustness and performance in contact-sensitive tasks.
	An Open-Source Robot for Learning Dexterity Manuel Wüthrich, Felix Widmaier, Felix Grimminger, Joel Akpo, Shruti Joshi, Vaibhav Agrawal, Bilal Hammoud, Majid Khadiv, Miroslav Bogdanovic, Vincent Berenz, Julian Viereck, Maximilien Naveau, Ludovic Righetti, Bernhard Schölkopf, Stefan Bauer CoRL, 2020 project page / PDF / arXiv / video An open-source robotic platform designed specifically for learning and evaluating dexterous manipulation skills.
	Learning to Explore in Motion and Interaction Tasks Miroslav Bogdanovic, Ludovic Righetti IROS, 2019 PDF / arXiv / video Continuous learning of exploration strategies through solving a curriculum of motion and interaction tasks of increasing complexity.
	Leveraging Contact Forces for Learning to Grasp Hamza Merzić, Miroslav Bogdanovic, Daniel Kappler, Ludovic Righetti, Jeannette Bohg ICRA, 2019 PDF / arXiv / video Using contact force information to improve robotic grasping performance through a learning-based approach.
	Deep Apprenticeship Learning for Playing Video Games Miroslav Bogdanovic, Dejan Markovikj, Misha Denil, Nando De Freitas AAAI Workshop, 2015 PDF Applying apprenticeship learning to training policies for playing Atari games based on expert demonstrations.

Website template by Jon Barron