Re-Balanced
In the pursuit of perfect balance and momentum
2017 GSAPP - Professor. Danil Nagy
Project Team Tigran Kostandyan, Zahid Nawaz Ajam⠀
Rebalanced is an exercise applying machine learning technology to the design process. The design is a subversion of the very same technology that it is using. Machine learning is traditionally used to optimize problems, while Rebalanced creates an unnecessary problem, and the algorithm is not allowed to solve it in the most efficient manner. The most optimal spinning top is an asymmetrical one. The problem posed to the algorithm is to create an asymmetrical spinning top. The algorithm receives an asymmetrical hexagon and attempts to correct the center of gravity, but it is not allowed to alter the shape of the original input. The adjustments are done through the addition and subtraction of spheres. The process through which the algorithm re-balances the spinning top ultimately becomes the formal expression of the design and an important part of its character.
01 Creating Asymmetry
A heptagonal design space was created within which the designer was allowed to tinker with the profile to produce the desired shape.
02 Correcting Dis-balance
Addition The genetic algorithm adds mass to the profile in order to balance the asymmetry.
03 Correcting Dis-balance
Subtraction The genetic algorithm subtracts mass to further refine the geometry for weight distribution and balance.
04 Re-balanced
The final outcome of the genetic algorithm is a balanced geometry which is able to spin well.
Fitness Criteria
To quantify the aspects of a good spinning top, three criteria were established. The primary factor, balance, determines if the top is able to spin. The second factor, mass distribution, determines how long the top is able to spin. More weight towards the exterior allows the top to spin longer. The third factor, weight, determines the handling of the top.
3D Printed Prototypes of Re-Balance
