🎓 — Семинар 12. Leveraging Single and Multi-Task RL Algorithms for Mobile-Aloha Robot | Aditya Narendra

Traditional Reinforcement Learning (RL) methods often require careful algorithm design, hyper-parameter tuning, and experimentation to perform optimally across multiple tasks. Multi-task models, however, offer increased efficiency, better generalization, and improved resource utilization, which are crucial for robots performing diverse autonomous tasks. On the other hand, single-task models often demonstrate better results and more robust task-specific policies.

In this paper, we demonstrate the versatility of these models through experiments on the Mobile Aloha robot, which has both manipulation and navigation capabilities. The main idea behind our work is to demonstrate the use of various types of RL algorithms (single and multi-task) for multi-control robots (in our case, Mobile Aloha) which has not been explored much in the past. The presentation would include the work done in two different simulators: Mujoco and Isaac-Sim.

👉🏻 Дата: 28.11.24, четверг в 17:00

📹 Трансляция Youtube

Подключайтесь к живому обсуждению и задавайте вопросы в прямом эфире! Ждем всех!

#семинары #robotics #RL