Robotics Institute Germany strongly represented at CoRL
The Conference on Robot Learning (CoRL) took place for the third time in Europe at the beginning of November in Munich. With over 1,100 participants, the conference, specializing in machine learning, grew by 20%. The organizers also recorded 33% more accepted papers than last year, with 264.
Partners of the Robotics Institute Germany (RIG) were strongly involved in the seventh Conference on Robot Learning. Six percent of the accepted publications came from RIG partners. They were involved in ten out of 13 workshops (partly as organizers), and the RIG network contributed six out of 27 demos. With more than 17 sponsors from industry, Google Deepmind and Meta as diamond sponsors, CORL has also become an important signpost for the industry. “Robot Learning has moved from being an esoteric topic to the most important topic in Artificial Intelligence-based robotics, essential for the economic future of an aging society. This increased importance is reflected by the outstanding work published at CoRL 2024.” Prof. Jan Peters of RIG-partner TU Darmstadt comments: “CoRL 2024 has become the essential venue for work on Artificial Intelligence-based robotics and is therefore crucial to the RIG. The publications by RIG members were very well received by the robot learning community.”
Trend: ?Imitation learning is becoming popular again“
Also, Majid Khadiv was satisfied: ‘We were able to show that we are a good host,’ says the professor from RIG coordinator Technical University of Munich, who was responsible for organizing the event as ‘Local Chair.’ A lot has also happened regarding topics: ‘Imitation learning is becoming popular again, but this time using teleoperation devices where a human shows the robot how to do the task using its embodiment,’ says Khadiv. According to the expert, researchers are currently taking shortcuts to teach robots complex movements by operating the robots remotely, using human operators. Although this is time-consuming and labor-intensive, it is probably the most promising way to teach robots complex manipulation skills. ‘Several companies also presented advanced humanoid platforms for research which could enable researchers to push the boundaries of what these robots can do in the near future,’ he added.
Robot of the future learns from experience
For Prof. Jan Peters, an intelligent robot that learns from experience and automatically adapts to its environment is the ?holy grail“ of artificial intelligence, robotics, and the cognitive sciences: ?That is precisely the goal that the CoRL participants are working towards. I am delighted that we can present our outstanding findings in two central areas of robot learning – the fine motor skills and locomotion of humanoid robots – in this innovative environment and thus contribute to future robotics.”
More information??
#CoRL #RoboticsInstituteGermany Bundesministerium für Bildung und Forschung
Professor at Computer Science Department, Leader of Interactive Robot Perception and Learning group
3 个月And another RIG member delivered an Early Career Keynote :) https://www.dhirubhai.net/posts/georgia-chalvatzaki_corl2024-robotics-ai-activity-7265122598563278848-fQHG?utm_source=share&utm_medium=member_ios