The bioloid robot climbs a freely configurable wall, this is NOT a predefinded motion sequence. It is part of an autonomous system that first looks for possible ways up (according to the abilities of the robot and to physical constraints) and after placing the robot at the start position of the chosen path the robots climbs up just by knowing the x/y-positions of the grips.
The final scenario will be a bioloid looking at the wall using an onboard vision system, detecting marked grips, calculating the best way up, walking to the start position and climbing up the wall, totally autonomously.
My part so far was developing a motion planning system which is a simulation of the real wall that calculates possible paths and developing a motion control system based on inverse kinematics so that now you can send a command like "left hand position 250/175" and the robot is hooking its new climbing hook-hand into a grip being at this position. (I made new hook-hands and climbing feet to make the robot able to climb, too)
See the other videos showing the accuracy of the inverse kinematics and explaining more details.
This is part of a university cooperation project done by Marco Wickrath.
(Technical University of Dortmund, Germany & University of Manitoba, Canada)