Hi,
I am researching on the impact of the implementation method on a vehicle’s dynamics. Therefore I have built the exact same ACC Modell in Simulink and implemented it via Simulink-CoSimulation, CarMaker PlugIn and FMU. Running the “SheerIn-Scenario” I have stumbled across certain differences that I am clueless about.
First of all: While the Plug-In and the FMU share quite the same simulation results, the Co-Simulation behaves differently. The velocity of the ego vehicle is different comparing the Simulink Co-Simulation with the other implementation techniques. this results in an increased distance(5m) to the Target Vehicle when it sheers in (t = 33s). Please note the velocity displayed in the added image. Especially the difference in the first 30s. The different brake
Secondly: There seems to be a bug with VC.Brake. If the FMU’s output to VC.Brake is <0.3, CarMaker is stuck in the Preparation-Phase when loading the simulation. The same bug happens when using the PlugIn. The only solution I found was to add an initial brake curve to the output of the ACC-Modell of the FMU and PlugIn. Therefore I can be sure that the initial VC.Brake value is >0.3 and the simulation starts workin. And yes, I have also added this additional Brake curve to the Simulink Co-Simulation. This is the first deceleration you can see in the image.
Additionally, the initial brake curve has wrong values using the FMU and the PlugIn. The initial brake that I added should start at 0.4 and slowly decrease to 0 at t =4s. However, only the Simulink Co-Simulation displays the values correctly when exporting the simulation results. The FMU and PlugIn both only have an initial Brake value of 0.284. To have comparable simulation results, have changed the initial Brake value of the Co-Simulation to 0.284 as well.
There seems to be a bug when using an FMU and CarMaker PlugIn, no? Do I miss something? Can anyone explain why the exact same models with the exacts same parameters have different simulation results? Which model is correct? This has a huge impact on validity of an ADAS simulation…
Kind regards 