Build a better GOM

Has anyone found a message that is sent just before the car is shut down? I am working on a DTT (Distance to Turtle) display mode and want to store some historical information to the eeProm just before the car is shut down to better estimate range when the car is powered back on.

I have to say that since starting this project I've developed a new respect for the GOM. We like to rag on how useless it is, but it is a very hard problem to solve and my own algorythm is starting to look a lot like the GOM (except it has the extra 5-6 miles the GOM hides). I'm thinking rather than measuring the actual mpkWh and using that, I will look at accelleration and braking style and use that to correlate to a number between 3 and 6 (hence the need to record historical information). Now if only I can find the canbus message that indicates what mood the driver is in...

I think the GOM should tick down with the odometer for the first 10 miles. If the guess is unfavorable move the display down max 1 mile for each half mile driven. If guess is favorable move the display down 1 mile after two miles driven. Never needs to increase unless charged.

Kills me how it moves down 10 then back up 5 within a couple miles.

A Ford researcher, Hai Yu, published a very interesting paper on the subject: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6183207
Yu's algorithm is able to distinguish, for example, the neighborhood road going downhill, from the long flat street without stoplights, from the Interstate highway, from the eternal traffic jam.

This paper presents a driving pattern recognition method based on trip segment clustering. Driving patterns categorize various driving behaviors that contain certain energy demand property in common. It can be applied to various applications including intelligent transportation, emission estimation, passive/active safety controls and energy management controls. In this paper, pattern features are first identified from high impact factors from static and quasi-static environmental and traffic information. A feature based trip/route partitioning algorithm is then developed based on data clustering methods. The driving patterns are finally recognized by synthesizing all partitioned feature zones along the trip/route where each partitioned road section is distinguished by an attribute of feature combination that will result in a distinctive drive energy demand property. The driving pattern recognition is a critical technology especially in solving problems like range estimation and energy consumption preplanning for the plug-in capable electrified vehicles.

Click to expand...

The hard part is getting reliable usable energy remaining.

I would average over a shorter rather than longer time,
so that the meter is "If I keep driving the way I am now,
my estimated remaining miles is = X."

Yes, that estimate will vary more than most folks would like,
but keeping the car at s steady speed and power consumption
for perhaps 10 seconds would give me a useful estimate.

To satisfy more people, make the meter average over N seconds,
from a few to days or even weeks.

In my Opinion, of course.

garygid said:

The hard part is getting reliable usable energy remaining.

Click to expand...

Yeah, given how I heard about the interesting gid values during the range test: both of fully charged cars and degrades ones once the battery was running low.

OP probably has already seen http://www.mynissanleaf.com/viewtopic.php?p=224711#p224711" onclick="window.open(this.href);return false;. I met Turbo3 at his house (he lives VERY close to me) w/surfingslovak and another guy. Turbo3 gave us a demo of the screens. I thought his way was a good/interesting idea.

cwerdna said:

garygid said:

The hard part is getting reliable usable energy remaining.

Click to expand...

Turbo3 gave us a demo of the screens. I thought his way was a good/interesting idea.

Click to expand...

It reportedly works really well. At least according to lemketron, who helped to design it. They calculate distance to turtle, distance to VLB and distance to LB. It's up to the driver to make the selection.

Nice. Lots of good inputs

smkettner said:

I think the GOM should tick down with the odometer for the first 10 miles. If the guess is unfavorable move the display down max 1 mile for each half mile driven. If guess is favorable move the display down 1 mile after two miles driven. Never needs to increase unless charged.

Kills me how it moves down 10 then back up 5 within a couple miles.

Click to expand...

Yeah, I think the existing one has a memory issue so you get that perturbation when you first start (my guess is it initializes a mpkWh register to zero instead of using the last known good value so it immediately becomes conservative until it has enough data to compute your actual driving conditions). Not sure I agree about never increasing, though. When I get low and concerned about making it home, I adjust my style and I think I would want to see the DTT meter reflect this.

walterbays said:

A Ford researcher, Hai Yu, published a very interesting paper on the subject: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6183207

Click to expand...

Thanks. Looks interesting and along the lines of what I was thinking yesterday. Something to read it on the plane today.

garygid said:

The hard part is getting reliable usable energy remaining.

Click to expand...

True, this may be mitigated by tracking your max gids on a full charge over time and averaging. We can use the pack volts=394V or SOC=80.0 on startup to determine when a charge was performed and compute a current_gid/average_gid scalar after each charge. On further thought, this may not work so well since we would have to average over a *year* to get rid of the seasonal variation.

garygid said:

I would average over a shorter rather than longer time,
so that the meter is "If I keep driving the way I am now,
my estimated remaining miles is = X."

Yes, that estimate will vary more than most folks would like,
but keeping the car at s steady speed and power consumption
for perhaps 10 seconds would give me a useful estimate.

To satisfy more people, make the meter average over N seconds,
from a few to days or even weeks.

In my Opinion, of course.

Click to expand...

I do have an "instant DTT" as well as the time averaged version (~15second averaging window). Instant was cool at first but eventually I decided the instant DTT was not very useful since it moved around from 0 to inf. quite quickly with subtle changes in throttle position and road grade. Made it very easy to see when there was the slight down or up grade in a seemingly flat road. The longer time constant gives something more meaningful but even that bounces around quite a bit (just more slowly).

surfingslovak said:

cwerdna said:

garygid said:

The hard part is getting reliable usable energy remaining.

Click to expand...

Turbo3 gave us a demo of the screens. I thought his way was a good/interesting idea.

Click to expand...

It reportedly works really well. At least according to lemketron, who helped to design it. They calculate distance to turtle, distance to VLB and distance to LB. It's up to the driver to make the selection.

Click to expand...

Yes, I agree. This looks like the best mechanism I've seen so far. In general I know about what mpkWh I can get based on the effort put into conservation and just having a value displayed based on 4.0 and 5.0 does bound most normal driving and allows me to make a good guess. None-the-less, if I can do it we should be able to program the computer to do it. Maybe a slick display would be one with a 4mpkWh DTT on the left and a 5mpkWh DTT on the right and a needle indicating how close the computer thinks you are to each (so if your current conditions are exactly 4.5, it will point to a spot exactly in the middle of the two numbers). I'm trying to see if we can make something as useful as Turbo3's display for non-geeks.

I like the way Tesla has two displays. One is a simple calculation using the SOC and the EPA 5 cycle consumption rate. It gives a very stable number that should be very close, but easily driveable if you are careful. The second number is SOC and the consumption over the past 30 miles, which is also fairly stable, but uses your current driving style. To do this you'd have to know the rate potentially over the course of several "segments". You'd have to keep a running calculation in your meter and not rely on the car for the recent consumption rate. You could also make it a bit more reactive by using 10 miles instead of 30. I have no idea how long the car uses for its average, but it seems like only a mile or two.

I am currently running with a moving average of about 4 hours and as such it is very stable (includes a couple of full discharge cycles). I have a work-around to the memory problem by just storing the state whenever I exit the DTT mode (pressing the red button). I just have to remember to cycle through the modes before turning the car off. Anyway I prefer my DTT to the GOM on the dash simply because it does remember my driving style and typical conditions so I get a realistic estimate when I get into the car. However, I had an interesting thought tonight. Instead of using a fixed time-constant for the averaging period, I am thinking of changing the time constant proportionally to the amount of remaining charge. When I get in the car I want to know how far I can go based on my usual driving habits (long time constant). But as I approach LBW I am more likely to try to stretch it (or get more spirited if I know I have charge to spare) and want a more responsive display that will show how far I can go based on more recent conditions. Maybe a simple formula: if X is how many kWh I have remaining, display how far I went on the previous aX kWh of energy; where a is a scalar.

surfingslovak said:

cwerdna said:

Turbo3 gave us a demo of the (WattsLeft) screens. I thought his way was a good/interesting idea.

Click to expand...

It reportedly works really well. At least according to lemketron, who helped to design it. They calculate distance to turtle, distance to VLB and distance to LB. It's up to the driver to make the selection.

Click to expand...

No amount of history or fancy calculations can possibly anticipate the driving that I'm about to do. Even if you add in elevation changes (which sounds great in theory), you can't anticipate traffic or what my driving style will be. My driving can average anywhere between 3 and 5 mi/kWh (for the SAME daily commute), depending on whether I'm late for a meeting (or maybe just wanting to humble some guy revving his engine at a stop light), or just taking my time.

This is why I have my WattsLeft set to display a range of DTT (distance to turtle) figures for efficiencies ranging from 3 to 5 mi/kWh. I always use Waze on my phone to know how far my destination is (and to pick the best route). If there's ever any question about whether I can "make it", all I need to do is reset the dash efficiency, and then adjust my driving style to keep that number above the mi/kWh needed according to WattsLeft. For me, since this basically tells me (assuming I'm on the freeway) whether I actually need to worry about how fast I'm going or not.

I think it all boils down to being able to see "best case / worst case" numbers. From there, having some sort of indicator ("needle" or otherwise, depending on the screen in use) that tells you (either instantly, or with some recent history, or maybe both) how close you are to either extreme could be helpful -- though somewhat difficult to implement in an add-on product. But that's really just a visual indication of what I already have by watching my dash efficiency and then looking to see where that falls (typically between 3 and 5) on the WattsLeft, which looks like this:



That image is exactly what I saw after driving to the October SF BayLeaf meeting in Union City last Saturday. As you can see, there's quite a spread between 43 and 73 miles to turtle, ranging from 3 to 5 mi/kWh. If I knew (for example) that I still needed to drive another 50 miles that day, I would have known that it was totally possible, but that I would likely want to average 4+ mi/kWh. 3 wouldn't cut it, and 3.5 would be cutting it (too) close. I prefer to see the range of numbers to make that determination, and rather than having the car (or any display) trying to guess how I might drive to make up a single DTE number to show me.

Yeah, that display is pretty slick but not what I am after. It does move in the right direction by eliminating the need to do math but you do still need to be "trained" in how your driving style translates to mpkWh. I am trying to improve on the GOM so that any stranger can get into your Leaf and get useful feedback on remaining distance. Rather then training the driver to the car I am trying to train the car to the driver. I cannot anticipate changes in driving habits but I can tell you how far you can go if you drive as usual. This is what Nissan attempted with the GOM but I think their big mistake was it appears to forget everything overnight I like the progressive sensitivity thing - seems intuitive to me. I'll code it up and see if it lives up to my expectation.

If I start up a long grade, or get on a freeway and achieve
a steady speed, in rain, wind, etc., then I would like to know
my range based on the past 30 seconds (or less) of driving, not
the past hours, days, or even minutes.

garygid said:

If I start up a long grade, or get on a freeway and achieve
a steady speed, in rain, wind, etc., then I would like to know
my range based on the past 30 seconds (or less) of driving, not
the past hours, days, or even minutes.

Click to expand...

I would just "reset the dash mi/kWh and look at WattsLeft".

But unless you're doing the Pike's Peak climb, or live on the hill and started climbing with a LBW or VLBW, why does it matter?

Wait, based on the past 30 seconds? Isn't that basically what the GOM tells you?

The GOM might use 5 or 10 minutes, or 5 or 10 miles.

I would like about a half-minute, or half mile "resolution".

Having the max and min DTE displayed on CANary is very useful - however it is not ideal. Presently it keeps track of my best and worst driving efficiency for each trip. Updating at the end of each trip if necessary. The problem is I really want it to only take round trips into account since there is a slight incline on my way to work (I need to use the total efficiency for the entire round trip from when the car leaves home to when it returns even if there are a few charges along the way). Sometimes I charge at work (if it is cold or I have lunchtime errands) so I can't just switch to the per-charge efficiency. I've failed to find any GPS data on the Carcan or Evcan (I would have to give up one of those use look at the Avcan). So far I have not come up with a reliable way to do this without GPS data so I thought I'd solicit ideas from the forum. Maybe there's something obvious I'm not thinking about. I don't want to have to tap the screen either (one nag screen is one too many). Some of the ideas I came up with for when I turn off the car and connect to the charger are:

• 1) Based on remaining charge (would work most of the time since I am below LWB if I don't charge at work but can't guarantee I won't be below LBW when I do).
  2) Based on time of day (could work except for days I work late and charge at work)
  3) Based on last 30 seconds driving pattern (steering, brake, accel, speed patterns) (could work but complicated and unlikely to be reliable - kid in the street, etc.)

cwerdna said:

garygid said:

The hard part is getting reliable usable energy remaining.

Click to expand...

Yeah, given how I heard about the interesting gid values during the range test: both of fully charged cars and degrades ones once the battery was running low.

OP probably has already seen http://www.mynissanleaf.com/viewtopic.php?p=224711#p224711" onclick="window.open(this.href);return false;. I met Turbo3 at his house (he lives VERY close to me) w/surfingslovak and another guy. Turbo3 gave us a demo of the screens. I thought his way was a good/interesting idea.

Click to expand...

ok then, I take it someone needs to update their profile?

I have the Lincomatic unit which has a Gid based display almost identical to this one and I can select LB, VLB, or T. I find this, in conjunction with the efficiency meter, to be very good in figuring out just how much range I have left. Of course, it still requires you to do a little mental adjustment based on your planned route...

lemketron said:

Click to expand...

Anyone have ideas on how to detect when a Leaf is at its home charger?

TickTock said:

Having the max and min DTE displayed on CANary is very useful - however it is not ideal. Presently it keeps track of my best and worst driving efficiency for each trip. Updating at the end of each trip if necessary. The problem is I really want it to only take round trips into account since there is a slight incline on my way to work (I need to use the total efficiency for the entire round trip from when the car leaves home to when it returns even if there are a few charges along the way). Sometimes I charge at work (if it is cold or I have lunchtime errands) so I can't just switch to the per-charge efficiency. I've failed to find any GPS data on the Carcan or Evcan (I would have to give up one of those use look at the Avcan). So far I have not come up with a reliable way to do this without GPS data so I thought I'd solicit ideas from the forum. Maybe there's something obvious I'm not thinking about. I don't want to have to tap the screen either (one nag screen is one too many). Some of the ideas I came up with for when I turn off the car and connect to the charger are:

• 1) Based on remaining charge (would work most of the time since I am below LWB if I don't charge at work but can't guarantee I won't be below LBW when I do).
  2) Based on time of day (could work except for days I work late and charge at work)
  3) Based on last 30 seconds driving pattern (steering, brake, accel, speed patterns) (could work but complicated and unlikely to be reliable - kid in the street, etc.)

Click to expand...

TickTock said:

Anyone have ideas on how to detect when a Leaf is at its home charger?

Click to expand...

i always assumed it went by the location you set as "home"

Yep, and you can set it to bitch at you via carwings email and/or text message after a user defined time limit if it has not been plugged in...

DaveinOlyWA said:

TickTock said:

Anyone have ideas on how to detect when a Leaf is at its home charger?

Click to expand...

i always assumed it went by the location you set as "home"

Click to expand...