Optimising a Jansen walker with genetic algorithms


Theo Jansen’s Strandbeest kinetic sculptures instantly inspire awe and wonder, and rightly so.  In describing the process used to design the original sculptures, he refers to using computational optimisation to derive the optimal dimensions for his linkages.  This is a problem I particularly enjoy: using software algorithms to model and solve  problems in the physical domain.

I therefore replicated a Jansen walker from scratch:  I use a physics engine to simulate the movement and forces in a Jansen linkage, this is plugged into a genetic algorithm that optimizes for a smooth gait with acceptable torque requirements.  Finally, I cut a set of parts to construct the physical implementation, thus crossing from the digital to the physical to test the final result.



The optimisation part of the problem itself is not particularly hard, as we are only dealing with 7 variables.  However, assessing the fitness of a particular instance of the linkage requires some enjoyable subtleties:  the joint needs to rotate smoothly, with as little torque as possible.  The toe of the linkage also needs to be as level as possible while on the ground, stay on the ground for the portion of the stride where it is carrying weight, and clear the ground sufficiently so as to avoid tripping.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s