Me and my colleague are working on a project using BeamNG.tech as part of an academic research initiative. Our goal is to replicate a radio-controlled toy car provided by our university, which will be equipped with sensors to implement autonomous driving algorithms.
Currently, we have successfully created a 1:1 scale replica of the car in the simulator (approximately 80x50 cm). The dynamic behavior of the simulated car closely matches the real model we aim to reproduce, which is fantastic. However, we have encountered an issue: the mass of the in-game model is an order of magnitude higher (~228 kg in the simulator compared to ~15 kg in reality).
We would like to significantly reduce the mass of the model to match the real car’s weight, but attempts to do so have led to instability in the simulation.
Could you advise us on an efficient method to reduce the model’s mass while maintaining its stability?
Thank you in advance for your insights and support!
Thank you for sharing your exciting project! Reducing the masses can be tricky but should be achievable. Here are some suggestions you could consider:
Scale Down Mass & Adjust Properties: When reducing the masses make sure you also adapt the stiffnesses and damping of the beams and torsion bars. The scaling might not be linear with the weight. You can try to maintain similar natural frequencies and damping ratios as the original car or you investigate in the stiffness and damping of your RC car.
Simplify the Model: Reduce the number of nodes, beams, or wheel rays (numRays parameter of the wheels). Note that fewer rays may make the vehicle’s ride “bumpy” especially at low speeds.
Investigate Instabilities: Use slow motion (ALT + ArrowKeys) can help to identify where issues arise.
Increase Physics Rate: If scaling down still causes instability, increase the physics step rate to 4000, 6000, or 8000 Hz (see BeamNG docu - physics fps). This is a quick and effective solution with fewer tradeoffs, but can lead to simulation times slower than real time.
Reaching a 15 kg mass is challenging but achievable with these methods. Let us know if you need more help! I would be happy to see your final result.