The estimated drive time (= score) for the SW Passage Route is 44 days.


This is about the same as the estimated time for the SW direct route, but it traverses much less violet terrain. It crosses mostly Blue terrain. Again the small patch of red on pavement terrain is probably overestimated. Due to Oppy's current position, this entry point is probably the easiest to access (might save a day).

This is my favorite route so far.

-Mike

And I'm likin' it a lot, JM.
But I could use some reassurance regarding that nasty splotch of red on the southerly leg. (Looks malignant.)
Are we able to focus in on that stretch at a higher resolution - say 5 meter, or even 2 meter blocks - to see if it's as perilous as it looks?
Finally, can you detail the procedure for inputting (golf term) the scientists' points of interest?
TIA

The W Spur route from Victoria Crater apron through the debris field to the fast green route intersection.
Black boxes are 50 m square:


Taking the most difficult terrain in the 50 m box to be the terrain type of the entire box (worst case scenario), here is the box-by-box terrain evaluation for the SW Passage route:

The estimated drive time (= score) for the SW Passage Route is a whopping 59 days.


There is a funky dark pattern with a brighter ring in the HiRise image near the bright green area. I'm flagging this as a potential Erebus-style dust trap. But the terrain surrounding this is pretty flat (aqua or blue) and going around this area shouldn't add to the drive time.

(There are a few of these S of the indicated path where the path bends to the E. The proposed path avoids these by a wide margin.)

-Mike

Yup! I'll put together some more detailed stuff covering the SW Passage.
(The proposed routes do not take into account any scientific targets (are there any at this point?), but the SW passage does just pass to the S of the 50 m diameter "cute" crater.)

Here's a comparison chart showing selected Victoria apron escape routes, with estimated times and terrain profiles for each of the routes:


-Mike

Mike, your work is just (sorry no enough good words come to mind).
Anyway, I just want to make sure here that you're aware that, apparently, the treck has started

My money is on the SW route.
It has the least amount of potentially dangerous stuff and the southerly leg gives us the half-pipes/pavement to drive between the bigger ripples.

Funny you mention it. The current drive direction mosaic is SW.

Anyway will be fun to see what happens...

STDEV analyzes of area south of victoria for southerly driving direction, analyzes was carried out at 1 mtr resolution, red spots showing potentially dangerous areas.

Unprocessed image of same area as above for comparison

HighRes STDEV analyzes of area SW of Victoria crater for SW driving direction, red areas state potentially dangerous places

Unprocessed image of above for comparison, had to reduce scale slightly to fit within attachments limits

Figured I'd have a go at implementing a route finding algorithm. The one I had mentioned earlier, A*.

Basically I took Juramike's red/blue/green map. (Thanks a million for your efforts by the way, amazing)
Broke it down into a grid with each grid square being X pixels wide/high.
For each grid location I average the pixels it contains and get the RGB value.
I break this down into the individual colour values:
R = 'Bad' terrain
B = 'Medium' terrain
G = 'Good' terrain

Then each grid is given a score based on these values, using a function I came up with from playing around. It's completely and utterly made up.
So the score/cost of travelling through a grid location is:


The route I have in the image here has grid squares of 5x5 pixels and took about 1 minute to generate on my 1.4Ghz Pentium M Processor with 500MB RAM so it's not too bad. The route shown is from the Top Left to the Bottom Right, just for illustration purposes.

If anyone has a good way of using the R,G,B values like above then let me know and I'll plug it in!

Edit: Seems I have a few bugs. God only knows how I got the route below from the code I just spotted.

Wow! Really nice.

I'm going to guess that your algorithm is trying to find a short path from the origin (0,0) which must be at upper left, to the largest X,Y which must be at lower right.

-Mike

Yep, forgot to mention that point about the origin. It seems the code is kinda doing what I thought after all so it's not too bad.

When it comes to balancing distance and traversability I'm not sure how it works out. I'll try at some stage to do something like you have to look at the type of terrain per step and see what the overall count is like.

Not tonight though.