Synchronization of Multiple Serially Actuated Robotic Legs Using Virtual Damping Control



Even though one-legged models have been found to be a useful

fundamental basis for understanding and controlling the dynamics of

running, animals and physical robots alike often use multiple legs

for additional support, dexterity, and stability. In general, the

dynamics of such multi-legged morphologies are more complex and

their control is more difficult. A common problem in this context is

to achieve a particular phase relationship between periodic oscillations

of different legs, resulting in different locomotory gaits. This

thesis focuses on a new method to achieve a desired synchronization

pattern across multiple legs, using series elastic actuation to

obtain virtually tunable damping coefficients of otherwise decoupled

and independent legs. In the first part of this thesis, leg models

are first considered separately from the synchronization

structure. Subsequently, we consider a Spring-mass-damper model as a

basic oscillator and investigate how the use of the damping

coefficient as a control input can enable multi-leg

synchronization. Following the investigation of this model, we then

proceed to consider the Spring-loaded Inverted Pendulum (SLIP)

model, which has been widely accepted in the literature as a

powerful tool to support the design of running robots, in order to

obtain a more robust and efficient control of relative phases of

different legs. Damping coefficients are, once again, used as

control inputs, using feedback from measured phase differences

between pairs of legs. We provide simulation results to support that

this is indeed an energy efficient way in which cyclic motions of

multiple legs in a system can be coordinated. Finally, the thesis

also introduces a physical platform design and construction that is

based on series-elastic actuation and is expected to support

experimental instantiations of the proposed synchronization

mechanisms for future work.





Date and Location

Date and Location