Jolt Vehicle Tuning Guide
I have tracked down the last regression that got introduced by the friction circle. Before each of the forces got evaluated alone so a lateral slip did not affect steering as much as with the introduced correct circle. Jolt’s vehicle controller just remaps the steering input. Both wheels are set to the same angle which as a result resulted in lateral slip on both wheels and as a conclusion also grip, even at very low speeds. It fortunately allows to set the angle per wheel separately, so adding a ackermann geometry calculation was quite easy. Stable cornering radius was reduced by around 20% just by that.
So i probably make a bullet list at some point and update the first post in this thread with the findings.
I was happy with the result, but i opened the rabbit hole of car simulation and as it turns out it is a very deep hole. Jolt’s vehicle is good enough to create racing games and with the simplified pacejka quite easily tunable to the gameplay wanted. Where it falls short is when you want to simulate modern/hybrid cars. So i found myself adding more and more code to the TirecallBack to simulate other non tire related stuff.
I run some experiments with using jolt as pure rigidbody solver and applying the forces manually to the chassis. The initial experiments have been so good that i started working on a SimCar controller. Since i already had the data from the 918 spyder i used that as reference for implementation since it actually has a very complex drivetrain. V8 trough gearbox, rear e-motor at the shaft, front e-motor at the axle.
Initial results look very good, without cheating on any of the real specifications i am very close to the actual reported numbers.
══════════════════════════════════════════════════════════════════
Porsche 918 Spyder — full 0→200→0 telemetry (Jolt physics)
Aero mode during accel: DRS (rear wing flat, drag-reduced)
══════════════════════════════════════════════════════════════════
── Time milestones ──
0 → 100 km/h: 2.90 s (real ~2.6 s, Δ +11.5%)
0 → 200 km/h: 7.65 s (real ~7.2 s, Δ +6.3%)
200 → 0 km/h: 4.13 s (real ~5.0 s, Δ -17.3%)
── Powertrain ──
V8 fuel burned: 9.79 MJ (0.31 L)
V8 peak power: 384 kW (515 hp)
Rear e-motor peak power: +209 kW
Front e-motor peak power: +71 kW
Battery drained (accel): ~1649 kJ (SoC 1.000 → 0.933)
Battery regen recovered: ~238 kJ (SoC 0.933 → 0.942)
Peak regen power: 60 kW
── Chassis dynamics ──
Peak forward G: +1.10 g
Peak brake G: +1.43 g
── Aero @ 200 km/h ──
Drag: 1691 N
Downforce rear: 1388 N
── Tyre wear / temperature ──
FL tread Δ=1.18 °C wear Δ=0.00001
FR tread Δ=1.18 °C wear Δ=0.00001
RL tread Δ=3.50 °C wear Δ=0.00002
RR tread Δ=3.50 °C wear Δ=0.00002
══════════════════════════════════════════════════════════════════
since i have no gpu currently, i was unfortunately not able to try driving the new controller. i am stuck on physics simulations
Discussion in the ATmosphere