Inspired by Kevin Lynagh, I maintain an ongoing list of ideas that I want to work on eventually, and possibly pair on. Some of them are inspirde by research papers that I want to replicate to better understand the intricacies, some of them come from unanswered questions that came up in my life somewhere. Other things are topics that I simply wondered about at some point, and want do dig deeper for a bit.

Note that most of these ideass are not projects I am supervising as part of my PhD, these are things I am doing for fun, and in general not very academic.


  • Optimal behavour of a vacuuming robot to cover the floor space efficiently: This would probably entail generating ‘realistic’ rooms automatically, deciding what sensors a vacuum robot should have (I am thinking only proximity), and what possible behaviours a robot can follow from that.
  • Landing a rocket: I am foremost interested in simply building a model (possibly in 2D for a start), and applying various approaches to
  • Examining the effect of modern stepsize controllers in numerical control
  • Exploring SAT solvers: Not ebtirely sure what I am specifically looking for here. Kevin Lynagh has a few ideas to the topic on his page, I am interested if this can be used for robotics applications.


  • Traveling Salesman Problem-Art: This involves approximating an image via a bunch of dots, and connecting them with the shortest path. Possible topics to pursue here are: optimal stippling, benchmarking different approaches of TSP-solvers, interfacing between python and cpp.
  • Simulating a leg through a pedal-stroke: Following up on my project I did before, I want to see when (and how) we actually apply pressure on the pedal.
  • Passing networks (PDF) in soccer over the curse of a season. How do passing networks change if the coach is replaced? Are there general trends visible in winning teams?
    There is an additional paper doing a deep dive of FC Barcelona in their most successful season here. Data for one season for several leagues is available here.


  • Moving horizon estimation: Enabling (amongst various other things) constrained state estimation, this is the counterpart to MPC for state estimation. This is a catch-it-all item on the list for various topics - please reach out if you are into state estimation/control.