Abstract: It is common to observe locomotion in animals and based on these observations develop hypotheses and models of locomotion. However, it is difficult and time comsuming to obtain observation data due to practical and ethical issues. In addition, verification of the developed models is difficult in particular when it comes to verifying the completeness of a model. LocoKit is an alternative tool researchers can use to understand locomotion. LocoKit allows researchers with no background in mechanics, electronics, or programming to build robotic models of locomotion. A robotic model complements the other approaches by being complete since there is no cheating in the real world.

At this tutorial you will a first be introduced to the LocoKit construction kit, to see what it is capable of doing. In the second half you will have the possibility to get to work with the system on your own. You will be getting a pre-configured quadruped robot build from LocoKit, on which you will have to optimize various parameters to get it to walk/run as fast as possible. In the end of the tutorial we will have a competition where the fastest robot wins.

To be part of the tutorial you will only need to bring your laptop and you will connect to the robot through a standard wifi connection. If anything else is needed we will provide it at the tutorial.

The LocoKit toolbox

Robot build from LocoKit

Abstract: The combined difficulties of hybrid dynamics and underactuation inherent in legged robots render the direct application of nonlinear controller synthesis tools to these systems a challenging task. To resolve complexity, the idea of task encoding through the enforcement of a lower-dimensional target dynamics (rather than through the prescription of a set of reference trajectories) has been employed in the relevant literature. Along these lines, the Hybrid Zero Dynamics (HZD) method was proposed as a systematic way to design control laws for legged robots with provable stability properties. In essence, this method "embeds" walking and running gaits in the dynamics of a legged robot by defining a set of (holonomic) output functions with the control objective being to drive these outputs to zero. The goal of this tutorial is to introduce the audience to the concepts that underlie Hybrid Zero Dynamics and its application to the control of walking and running robots. A broad range of models will be considered, and theoretical aspects as well as implementation details will be discussed.

You can download the MATLAB files here:

You can download the presentation here:

• View
• Edit
• History
• Print
• Backlinks