Project: 3D Driving Model

[noeryan]

Hey guys!

I'm relatively new to EcoModder.com

I'm a student of Aerospace Engineering and have an idea that really doesn't encompass much more than Physics.

Ok so here we go.

I would like to develop a short program which provides driving information to accurately throttle the engine for optimum fuel efficiency.

The program would take into account:

Topography of Route
Vehicle Characteristics (Drag, Rolling Resistance, HP)
Route Characteristics (Stop Signs, Traffic Lights, Turns, etc.)

This would be used for routes that are normally driven by the user; which I believe are the majority of one's driving. From this knowledge of route, accurate distances (mapquest.com, google maps, etc.) can be calculated between each Route Marker (stop signs, etc.) along with a topographical map study to determine the gradient of the route (since most topo maps use 10 foot intervals, it would be fairly accurate).

Using all of this data, we could determine the amount of kinetic energy that would be needed to move the vehicle with its particular parameters the distance of the route with some limits of min and max velocity for the route.

Without explaining all of the details of the calculations...

I'd like the OUTPUT to be set to tell the user what velocity he/she should be driving at certain key "Checkpoints" along the route to ensure they are maintaining the proper velocity profile.

For example:
I have a route to school which has a series of hills of different heights and four red lights. I would input this route into the program and maybe even an online database. The program runs the analysis on the route and delivers the average MPG I should receive and distances in which I should be at certain velocities.

Maybe the Output at first could look like "Accelerate to 45 MPH until reaching the bottom of Hill #1. Consistently decelerate up Hill #2 until the top is reached at 35 MPH.."
Eventually I'd like to see a display that gives up and down arrows to tell the driver how far to push on the throttle... Maybe it could even get produced into its own cruise control which adjusts the throttle by itself just by uploading the route and having a GPS guided processor letting it know where in the route it is.

This is a long way down the road. The first attempt would just to get the algorithms to accurately produce data with simple routes.

I could explain more thought on the matter but I would like to know if anyone would be interested in helping with the project before I waste my finger strength typing out more details.

Thanks guys!

[Piwoslaw]

Good idea

(1)- Knowing where traffic lights are isn't enough, you should also know what color it is and when it will change.

(2)- The program should not only know what kind of congestion is on which route, but should also "see" more locally, i.e. how many cars are in front of mine and what they're doing. Some kind of radar/distance device up front would help.

(3)- The program should be adaptive. It should see how I'm driving a certain route and confront it with what it was suggesting. After a few goes it would start to learn and adapt to how I drive on that route.

[daisy]

Sounds fascinating. You may be interested in Karen Pease's EV charge calculator, and my own project to make kinetic energy easier to visualize.

Of course, Karen's work is more towards a simulator, not a driving guide. You'd need to fill in all the specifics in Karen's simulator, of course. I don't know what assumptions it makes about throttle, but they're obviously different for an EV than an ICEV. A brake-specific horsepower graph for your vehicle's engine would help tremendously.

[noeryan]

Thanks for the encouragement.

Piwoslaw, my idea for the project would to give you the IDEAL velocity profile for the trip. If congestion gets in the way of the route, its just part of the inefficiencies of the trip.

I tend to get in way over my head with projects that are too detailed for me to accomplish... I want to start simple with simple outcomes.

I've got some MATLAB experience, but it may be hard to get your hands on it if anyone wants to help out so I'm going to try to just use Excel and for now try to keep it open source so people can use it.

I'm gonna list the algorithms in the thread so people can correct me when I screw it up

[TestDrive]

Originally Posted by noeryan

Piwoslaw, my idea for the project would to give you the IDEAL velocity profile for the trip.

Is that for DWL, P&G (EOC), Pulse and Neutral Coast with Engine On or some other FE driving stategy?

[noeryan]

Testdrive: I've never used Pulse and Glide. However, with the terrain, it may be a slight form of it.

This would only entail smooth accelerations I believe...

ALSO **IMPORTANT**

Thinking of a way to do the calculations and since it will be a trial and error type of thing (backwards differential equations), I may have to do this in MATLAB.

We could use Excel but I believe the intensity of the program with its MANY differential equations might slow down the calculations and we would be limited to the number of cells etc.

Maybe we can start with excel to get the algorithms correct and then I'll switch it to MATLAB to put more depth into the program.

I would be using Euler's method of differentiating if anyone was curious as to the set up. Also if anyone has a more computer friendly algorithm please let me know! Since I would only know the start and end conditions (being stopped) and the redlight/stop sign conditions, I would have to do the differential equations numerous times in a loop until I came across one that gave results that I liked... this would be easy but takes time for the computer to run (not real long... really)...

[Bicycle Bob]

I'd be inclined to put a GPS and a laptop together, and have it gather most of the route data from a first run. That could also be a useful base data point for the customer. The driver would confirm that progress was being blocked by traffic when present, although soon a camera might help with that, too. (Police cars can automatically check every license plate visible; spotting cars should be easy.) We might also key in, speak, or use the camera to get timing data on upcoming traffic lights. The program might need help guessing the realistic speed options on a given stretch, or might just go by experience as history builds up.

Perhaps the output could be audio coaching on when to coast, speeds to drive, etc, like a rally navigator on an economy run. An accelerometer would probably improve on the data available from GPS, especially for grades. An interface with the vehicle computer might reap valuable insight, too.

[noeryan]

I'm glad to see interest!!!

As of now, I'm just going to get my bearings on how to set up the initial algorithms and do maybe an incline then decline to test them out. I can't figure out how to reverse the differential equations (more unknowns than equations...) unless I can somehow use the initial conditions and the final conditions to give me enough knowns. Using this would mean I had to make the equations more flexible to account for every hill all in one set instead of breaking the trip into mini trips and doing a set of equations for each linear gradient.

I like where you are going with the GPS/laptop config Bicycle Bob. If you know much about this it would be awesome.

I'm going to stick with the math and very simple models for now and get something producing some data. From there I plan on expanding the trip and testing it and debugging. Next would be to add in gear shifting (try to stay in 5th gear velocities and power requirements)... its a step by step thing. For the first step I'll create the Euler Equations in Excel and post it for you guy's expert review

[dcb]

I would suggest targeting a complete "vertical slice", so you can get something working and see if you are in the ballpark. Limit the frills (i.e. tire heatup and leak down).

At some point you have to "reverse engineer" the cars CPU as well, to know what conditions it will go into enrichment, but you might just call it stoichiometric for now, and assume standard day conditions.

But yah, every time I see a PC steering wheel and pedals at a garage sale I want to build a FE driving simulator around them

[Bicycle Bob]

Gosh, DCB, I wonder if there should be a car-race game with limited fuel. . . Winner uses gas where it helps, and coasts a lot. Did you know that if you run experienced cyclists and beginners through the same course and time, the regular commuters use HALF the oxygen. The game could have visual feedbacks to coach the same improvement.

Hi Alabama, I'm no expert on programming interfaces, I just know what I'd like to deal with. People seem to be pretty happy with GPS units that give synthesized voice prompts for turns en route, so this could fit as an added feature. Maybe you can plug a GPS into a laptop and see what sort of software shows up in open-source to run them, and then add features.

(I did do the calculations for a streamlined HPV on a flat, serpentine course once, with hard inputs - it is the obvious approach, but not very user-friendly. I wrote and used it while riding in a car to the race. The training seemed to be working, until I fell off the pace badly, and noticed that I had a flat tire. Anomalies abound.)