Autonomous state-based flipper control for articulated tracked robots in urban environments.
T Azayev, K Zimmerman - IEEE Robotics and Automation Letters, 2022
Blind hexapod locomotion in complex terrain with gait adaptation using deep reinforcement learning and classification.
T Azayev, K Zimmerman - Journal of Intelligent & Robotic Systems, 2020
System for multi-robotic exploration of underground environments CTU-CRAS-NORLAB in the DARPA Subterranean Challenge.
T Rouček, M Pecka, P Čížek, T Petříček, J Bayer, V Šalanský, T Azayev, ... Accepted in Field Robotics 2021
Previous work
Bachelor's thesis: Object Detection in High Resolution Satellite Images
In 2020 and 2021 I participated in the urban and final rounds of the Darpa subt challenge in the USA. During the preparation and participation I contributed to a variety of tasks such as articulated tracked robot control, system integration in ROS and tuning of mapping and planning algorithms, payload design and assembly, SPOT robot integration and control.
Interests and research
I have studied and applied deep learning methods for various robotic platforms, such as hexapods, quadrotors and 1/10th scale car platforms both in simulation and on the real platform. My current interest are sim-to-real transfer, few shot learning for robotics and differentiable optimization with focus on control. I am also interested in non deep learning based high performance system control methods such as MPC and iLQR.