generative design


Research activity / master projects at ROBIN IFI UIO

Generative design

  • Using various optimization algorithms from AI to automatically generate shapes that perform in an optimal way with respect to specified constraints/external forces. Physics simulation is used in the optimization loop.


  • Evolving physical shapes based on simple growth algorithms. Example.mp4


current work

master projects


  1. Attractors
  2. Holes/erosion
  3. Local growth
  4. Reality gap
  5. Uniform wrinkles in the overall mesh
  6. Printing in real 3D by using the Z axis while printing (no STL slicing)
  7. Growing the whole geometry between two joints from a single edge
  8. True 3D printing (no Z slicing) of wall pattern in Z direction
  9. Growth of solar cell based structure with orientation against the sun

Possible projects (max 2 students)

A typical master theisis work flow

The following projects are based on fast, thin wall / spiral FDM printing, modeled and simulated as a 2D mesh. Our Ultimaker 3D printers will be our primary printers, but also other printers can be used when needed

The optimization goal is to automatically obtain shapes that optimizes the strength/weight ratio of the printed part. Strength simulation are either done based on simple models/algorithms that that are included in the overall optimization program or can be done by standard FEM library components if preferred.

  1. BASIC SHAPE optimization using attractors in 3D space. The optimization parameters will be the xyz positions of the attractors.
  2. LOCAL GROWTH by demand. By simulating the shape displacement/strain of the part when exposed to external forces, local shape modifications (growth/wrinkles) can be added to areas where the displacement is found to be at the maximum. In this way more material is added only to areas where it is needed optimizing the strength/weight ratio of the part
  3. HOLES/EROSION optimization. Areas in the mesh where the simulator finds uncritical for the overall strength, can be removed or replaced with holes. The location/size/shape of these holes can be optimized.
  4. WRINKLE optimization. By applying uniform wrinkles or corrugated waves to the whole mesh, the strength/weight ratio can be optimized. Modeling and optimizing the shape of these wrinkles will be the focus in this project
  5. REALITY GAP minimization by automatic parameter optimization. Our simple physical mesh model is based on a small number of constant parameters. The values of the these parameters can be optimized by generating and using a dataset from our physical test rig where various strength properties of a printed thin wall FDM sheet is measured


  • C++ / Processing (Java) / Blender, AutodeskDreamcathcer/Netfabb, Ultimaker, physical test rig