Josua Spisak
Hamburg University
The capabilities of humanoid robots are continuously improving, with new skills ranging from dancing to pouring being constantly added. While these abilities can be highly impressive, the teaching methods often remain inefficient. To enhance the process of teaching robots, we propose leveraging a mechanism widely observed in the natural world: Learning from demonstrations. On this topic, we will analyze the challenges of robots imitating humans and propose two approaches to address these challenges. First, we will focus on how the demonstrations can be perceived and transferred to the robot’s perspective to allow for a simpler imitation. The second approach goes further and directly uses the demonstrations to generate joint values that imitate the demonstrated pose for the robot. In both approaches a key aspect is how to transfer what is seen in others to the self and we demonstrate progress on our humanoid robot NICOL (https://www.inf.unihamburg.de/en/inst/ab/wtm/research/neurobotics/nicol.html)
DFG SPP "The Active Self" 