Daniel Renjewski (PhD candidate 2008-2012)
ResearcherID: A-1371-2009
e-Mail: daniel.renjewski@uni-jena.de
Since centuries biological locomotion is a technically unrivaled archetype for many kinds of machines. Still today the biomechanics and control of human and animal legged locomotion is far from being well understood.
My research focuses on translating mechanisms found in biomechanical experiments and computer models into technical models to challenge theories on robustness, stability and control of locomotion in nature. As technical models existing robots, i.e. JenaWalker series and RunBot, are used as well as new systems are developed.
In future, the developed mechanisms will be integrated in robots and biomedical devices, such as orthoses and prostheses, to validate mechanisms experimentally and to push beyond the state-of-the-art of current mobility supporting systems.
Currently I am supervising the bachelor's thesis of Robert Schirmer. We investigate the influence of hip paramaters on forward speed. Therefor we optimise a multibody robot model (see below) for maximum forward speed. An initial mapping gave us an idea, where the model is capable of showing continous gait. The mapping results are indicated in gray, the donut size symbolises the forward speed. Animated in colour are successful iterations of a particle-swarm-optimisation algorithm applied on a non-continous function.
An extensive scan of the current version of the robot's computer model regarding the hip parameters anterior extreme angle (AEA), posterior extreme angle (PEA) and hip extension voltage has been done using the university's compute server Meggy. A number of parameter sets lead to successful walking, but gaits differ substantially.
An extensive scan of the current version of the robot's computer model regarding the hip parameters anterior extreme angle (AEA), posterior extreme angle (PEA) and hip extension voltage has been done using the university's compute server Meggy. A number of parameter sets lead to successful walking, but gaits differ substantially.
The pilot experiment unveiled that the choosen way to acquire data leads to not entirely reliable data. I changed the setup and in cooperation with Moritz Maus an experiment was conducted. The robot walked on a circular even track and 8 infrared cameras captured markers on the robot. Thus kinematic data over more than 1 minute and about 130 steps with much higher precision could be recorded.
In a pilot project kinematic data of a biped robot were acquired using a high-speed camera (600fps) and tracking LEDs.