@ ntxdave: Thank You! The plane is completely modelled in SketchUp with SketchyPhysics (plugin).

And here is a short solution for a path following plane. I think it is a bit complicated and that there are simplier solutions for a calculation of the direction vector. If anyone has ideas ... I'd like to replace the forward flying cube and the LookAt function by a Rubi-formula in the script of the plane.
The model is a simple double-decker from the warehouse without Physics. I have it relatively heavily rebuilt and have added physical properties. The scripts are used from earlier SketchyPhysics posts about camera curves and path following object by Mr.K. and other authors and are also slightly modified.

how it works

After the first test flight comes now the first start ... Despite some obstacles that seems to work quite well.

first start

My goal is to test in a small rural "world" different ways of animation with SketchyPhysics
and to knit a short story. Started with experiments on Physics-waves of Anton S., further
experiments were carried out with emitters - as an example of chimney smoke, which could not be recorded normally (except with screen recorders). I start the post with a new test with the aim of reproduce a moving water surface.
The relaxed flight missing a bit of content. This should, just like the environment model,
develop a little further.
The tests are carried out even with SketchUp Pro 8, SketchyPhysics 3.5.6 and Twilight renderer v.1.5.4, as these plugins converge very round. Later I hope to be able to transfer the whole thing to a current SketchUp version and to MS-Physics.
If anyone has ideas on content for the story or for Physics tests, then post it here.
Here is a first short flight over the lake. Since only files can be appended with a maximum
size of 2 MB, the videos are heavily compressed and scaled down.

first short flight

As pilou mentioned - SketchyPhysics - with unexpected results... not really for natural stonewalls. But it's worth to try...

Splittest 2 - unexpected

I could take on the assignment of physical properties with SketchyPhysics. I would suggest the use of a type inverse kinematics. The tricycle follows a path and all movements lead to the driver from it. The whole thing could then be followed in a WIP under SketchyPhysics.

Here some later happy Easter greetings to the friends of SketchyPhysics. Have a little too much tested and tasted the "Easter water"... Also a thank you to Anton for the important improvements to the plugin. One can already produce a lot of funny things with it without too much bugsplats. The video is a salute to my 2 week old granddaughter Matilda.

Export tested with a simple example - Bug Splat after export of 2 DAE-files ... The script is in the table. Have no idea. I am a beginner in programming... Export tests in the SKP or image data format works well in this way. Sorry ...
The better way is perhaps an enhanced Sketchy Physics Replay Collada export function. I recommend Anton S. for this.

.dae direct export test

The demonstrations and examples are impressive and inspiring. This offers completely new possibilities to deform objects specifically and flowing. "Fredo6_FredoScale Radial Bending" on the next level.
Will test it under SU8.

Your new script runs great! Haven't seen your full post this morning. Don't know why. Thank you so much! Now I can implement camera paths or walking lines for moving objects without conflicts.

bones ...
Sketchy Physics models from 3D-warehouse

"Physics-bones" from 3D-warehouse

Many thanks for the quick response! Meanwhile I had already found the small screws to adjust the parameters. The problem of the influence of other free geometry during physics simulation remains. Here the intermediate state of my Easter physics experimental scene:

Easter wave test

I will not slow your enthusiasm, but I think TIG has a plugin for it: CADup V1.2a. Whether it is suitable for the newer versions of SketchUp, don't know ...

Hi, Anton. A small request: Is there a way to change your script so that other non grouped entities (lines, curves, faces) in the model are not influenced? One possibility might be to put the geometry of the wave surface on a specific layer and to drive only these. When attempting camera paths (curves, polylines) to use in the wave model, these are extremely manipulated and broken after the start. Thanks in advance!

Experiment with indigo renderer (focal point, aperture).
Motion blur with copies and different transparency.

Could not resist - SketchyPhysics & Twilight.

Here a quick nail emitter. You have to install SketchyPhysics before. Not a new thing but always funny. Pause Physics on the desired position an save the SketchUp file as a copy an you will get the pile of nail.

example nail emitter

nail emitter

Hi Anton. Here's a quick question, I have reduced the water surface to an almost realistic size. The bowl is about 1 m in diameter. How can I change amplitude, wavelength and speed of the waves so that it fits again. Thanks in advance!
Yet another problem: "drops" of about 2.5 cm in diameter do not work in the normal setting worldscale 9.0 as an emitter. Here I have to reduce the worldscale to 7.0. Is this normal?

wave test

So, I agree 100%. The implementation in Physics would be a big step. As a plugin to prepare concave models for physics it would be very useful.

Could be Faces To Components by HDPV (S4U - Huynh Duong Phuong Vi)...

increase wave travel distance and reflection on objects would be great...