First ever calculation of the symbolic non-linear equations of motion of the bicycle with open source software
by Jason Moore — last modified Aug 06, 2011 01:20 PM
Gilbert's made enough progress on sympy.physics.mechanics to be able to calculate the bicycle equations.
Through two Google Summer of Code projects Luke Peterson (2009) and Gilbert Gede (2011) have developed an open source piece of software that can symbolically form the equations of motion for complex dynamics systems. The software, a mechanics module in sympy, is still in development but will be merged into the main Sympy software and bigger packages like the Enthought distribution and Sage later this year.
I wanted to share Gilbert's blog post about correctly deriving the symbolic non-linear equations of motion of the Whipple Bicycle Model and subsequently linearizing the equations symbolically with the software. This is a big deal for us, as software capable of this is few and far between and as far as we know this is the first free open source version that can handle complex dynamics symbolically.
So if you can't get your hands on software like AutoSim (BikeSim), Autolev, FastBike and other proprietary options, there is now something available to anyone that can derive the equations of the bike/motorcycle etc or any other equally complex dynamic system.
The incubus for this project is rooted in the some of the frustrations we've felt regarding being able to use other people's code and models from proprietary software packages. To us, the open source model fits really well with academia where we rely on the ability to stand on the shoulders of giants. We hope that tools like this allow for the reuse of research code it much broader fashion. For example, journal papers publish software code that is for software that is not necessarily attainable by reader, which renders it fundamentally useless. That's certainly one of the reasons we went to the drawing board and derived the equations of motion from scratch even though hundreds had done it before us. It's not that this is necessarily a bad thing, but I'd be nice to be able to use models more easily if you don't have the desire to derive them yourself or actually run and check code to see if you think the equations it produces are actually correct. The current state of affairs doesn't necessarily let you do this.
Congrats to Gilbert and Luke and you can follow Gilbert's blog about the progress of the software.