You are here: Home Teaching Reading Groups Autonomy for Manipulation

Autonomy for Manipulation

The Robot Learning Lab organizes a monthly reading group (no ECTS) on Autonomy for Manipulation in cooperation with the Autonomous Intelligent Systems Lab. Here we discuss recent publications on the field as well as work in progress in our groups. The main focus of this reading group is the area of autonomous robot learning, control and manipulation. Speakers are the PhD students of the AIS and RL labs.

During the meeting, the presenter gives a short talk (25 min) summarizing a paper either previously voted for or their own research work and leads a subsequent discussion (30 min). The target audience consists of Professors, PostDocs, PhDs, and interested Master's students.

To stay informed about all future instances, sign up for the mailing list.

rg_manipulation


Next Meeting:

  • Date: 20/05/2022, 3pm
  • Presenter:Eugenio Chisari
  • Title: Correct Me if I am Wrong: Interactive Learning for Robotic Manipulation
  • Abstract: Learning to solve complex manipulation tasks from visual observations is a dominant challenge for real-world robot learning. Deep reinforcement learning algorithms have recently demonstrated impressive results, although they still require an impractical amount of time-consuming trial-and-error iterations. In this work, we consider the promising alternative paradigm of interactive learning where a human teacher provides feedback to the policy during execution, as opposed to imitation learning where a pre-collected dataset of perfect demonstrations is used. Our proposed CEILing (Corrective and Evaluative Interactive Learning) framework combines both corrective and evaluative feedback from the teacher to train a stochastic policy in an asynchronous manner, and employs a dedicated mechanism to trade off human corrections with the robot’s own experience. We present results obtained with our framework in extensive simulation and real-world experiments that demonstrate that CEILing can effectively solve complex robot manipulation tasks directly from raw images in less than one hour of real-world training. .
  • Zoom Link

Paper Voting:

Please provide your vote regarding the paper to be presented during the subsequent meeting:


In case you would like to individually suggest a paper or topic to be presented: Suggestion Form

Future Dates:

Date Name (Lab)
20/05/2022 Eugenio Chisari
17/06/2022 Oier Mees