GregLocock
Automotive
- Apr 10, 2001
- 23,701
I've been asked to look at how to generate 3d road profiles for use in driving simulators and ABS performance evaluations. Luckily I have some real road profile data, measured every 6.35 mm, for each wheeltrack. I have also spent a fair while on a literature search.
Terminology : x is distance along the track measured from the start of the road, y is across the track measured from the centre of the road,left is negative, z(x,y) is the height at that point. wn=wavenumber (2*pi/wavelength in m). Coherence is (roughly) the proportion of the energy at each frequency in the second track that is due to the first signal. More accurately it looks at how stable the phase is between the two at each frequency.
An older approach is to assume that the road surface is isostatic, that is, the psd and the plot of coherence vs wavenumber between adjacent tracks is the same across the road as along it for a given spacing between slices . Bogsjo's excellent thesis suggests this is a bad idea. p101 onwards This doesn't exactly amaze me, given that roads are made using rollers that drive along the road, not across it.
Obviously most work uses the coherence between the left and right track, but for the visual part of the job I need a grid size of say 10mm. I am hoping that the coherence is an emergent property, that is, the coherence at a separation of 1600mm will be the cumulative coherence of 160 adjacent 'slices' 10mm apart (my brain will explode if that is not right).
Does anybody have any direct, or at least relevant, experience with synthesising road profiles? I'm floundering a bit as the signal processing side is beyond my usual level (I think I can do it but it's easier to ask!).
.
Relevant previous thread
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376
Terminology : x is distance along the track measured from the start of the road, y is across the track measured from the centre of the road,left is negative, z(x,y) is the height at that point. wn=wavenumber (2*pi/wavelength in m). Coherence is (roughly) the proportion of the energy at each frequency in the second track that is due to the first signal. More accurately it looks at how stable the phase is between the two at each frequency.
An older approach is to assume that the road surface is isostatic, that is, the psd and the plot of coherence vs wavenumber between adjacent tracks is the same across the road as along it for a given spacing between slices . Bogsjo's excellent thesis suggests this is a bad idea. p101 onwards This doesn't exactly amaze me, given that roads are made using rollers that drive along the road, not across it.
Obviously most work uses the coherence between the left and right track, but for the visual part of the job I need a grid size of say 10mm. I am hoping that the coherence is an emergent property, that is, the coherence at a separation of 1600mm will be the cumulative coherence of 160 adjacent 'slices' 10mm apart (my brain will explode if that is not right).
Does anybody have any direct, or at least relevant, experience with synthesising road profiles? I'm floundering a bit as the signal processing side is beyond my usual level (I think I can do it but it's easier to ask!).
.
Relevant previous thread
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376