Time-optimal trajectory for a race car

[roborace]

Hi everyone,

 

I currently work on the minimum lap time problem (time-optimal trajectory) for an autonomous electric race car. (using a single-track car model) 

There are several papers about this topic. The authors use GPOPS-II, ICLOCS2, PINS or Falcon.m in order to solve the OCP.

Unfortunately, these tools have no interface to python, which I need!

1) Do you know if there is an suitable optimal control software for python, which can handle this kind of problem (pyomo, gekko...)?

2) Can you recommend casADi for this task? 

3) What's about the speed and robustness in comparison to other free/commercial Optimal Control Softwares?

4) Do you recommend multiple shooting or direct collocation for this problem?

5) Is there an example for casADi with lookup tables oder complex path constraints?

Thank’s a lot for your help!

[Joris Gillis]

Hi,

CasADi doesn't offer a plug-and-play OCP solver, but rather a couple of fast building blocks that you can fit together to do OCPs.

We list a couple of OCP packages built on CasADi at the bottom of our main page casadi.org

I'm not sure how efficient they are, but it is possible to achieve state-of-the-art performance if you combine CasADi with a good mathematical formulation and a good solver.

I'd think a race car would involve non-stiff nonlinear equations which is good for a collocation method. Collocation also typically has a larger region of convergence than multiple shooting.

We do support lookup tables ( https://web.casadi.org/docs/#using-lookup-tables ) in ODE or in constraints. I'm not sure what you mean with complex constraints. 

Best,

   Joris