A Joint Motion Model for Human-Like Robot-Human Handover


In the near future robots will encounter
us more and more in everyday life: Besides tasks that
the robot can solve autonomously there are also many interactions with a human being. One of these tasks is the transfer of objects between robot and human.
This results in close contact between man and robot. Which can cause insecurity
or anxiety especially among inexperienced people. In order to
increase the feeling of safety, we have developed and tested a movement model
that is based on the movements of humans. The imitation of human movements is a
well-known method for improving the feeling of safety . Models for human-like
transfers developed so far focused primarily on a human-like trajectory. However, this can lead to inhuman robot poses, like these. That is why we have
developed a model that is oriented on the joints and not on the pose of the
end effector the path poses are thus in the joint space and not in the Cartesian
space in order to determine joint movements we analyzed visual data from
two studies in these studies two people handed over
an object with the help of various analyzes movement functions for the
different joints were developed during the analysis and development we noticed
two different variants in the first variant the elbow moves
continuously into the target position in the second variant a more pronounced
flexion of the elbow at the beginning of the movement requires an additional
extension at the end of the movement the variant can have different degrees of
intensity after the analysis we transferred the movements to two robots
a humanoid pepper robot and a cucurbit arm
in order to test the performance and the sense of safety we then carried out
another study we asked 15 participants to take a look at the robots all
participants were secretly divided into two groups ten people formed the
experimental group the remaining five are the control group the study was
divided into two phases to see whether the subjects could perceive a difference
in the movements in phase one the subjects were not informed about the
details of the experiment both groups were shown two movements of each robot
the humanoid robot and the traditional industrial manipulator the movements
were different for the groups the experimental group was shown a
human-like and a non human-like movement however the control group was shown the
nonhumanoid movement twice per robot the first phase was followed by a
questionnaire survey questions regarding the recognition of differences and human
similarity had to be answered afterwards an explanation was given to
the test persons they were informed that they were shown two different movements
and that one of them is more human-like there was no explanation of what the
differences are nor what the more human-like movement is in phase two the
movements for both robots were shown again the different movements for the
experimental group in the same movements for the control group the questionnaire
was then repeated with an additional question regarding the feeling of safety
the statistical analysis of the data of the study shows different results
details can be found in the paper in summary the subjects were able to
perceive the difference between the movements this was already clearly
achieved in phase one of the industrial robot with the humanoid robot the
subjects only noticed the difference in phase two
after the explanation this is probably due to the small difference between the
motion models of the humanoid robot the control group ruled out that this change
was based only on the explanation in comparison it can be seen that our joint
movement model is regarded as more human-like here – the difference is
clearer for the industrial arm our model also performs better in terms of safety
further results are presented in the paper you

, ,

Post navigation

Leave a Reply

Your email address will not be published. Required fields are marked *