Adna’s paper is not fully suitable for the journal she submitted the paper to and is now resubmitting it to a different journal.
Maze experiment is working on simulation and on the robot. Adna used images to extract the shape of the maze. In addition the wire maze is explored with haptic feedback – 5 trajectories where recorded using human demonstrations to learn the shape of the maze. During the demonstrations the loop was touching the wire in the top/left/bottom/right side of the loop. Next step is to collect human data solving the wire maze. Changed the plan for the human experiment to use a handle instead of the glove.

First Robot experiment with human demonstration will take around 90 min per person including demonstration and training. Experiments will start in ~ October. Second robot experiment (using PlayStation controller), explores online adaptation. Instead of having a fixed percentage of autonomy Ege wants to give a +/- 10% range around offline adaptation points to get better data to learn a regression model – data is to volatile for regression models. Final decisions on the experiment design in the next 2 weeks. 2nd Experiment will take 60-90 min and start around 2 weeks after Adna’s experiments.