Autonomous Robot Taxi Obstacle Detection, Passenger Logic, and Hardware Module Suggestions

Hello,
I am working on a * Autonomous robotic Taxi*..that would track passengers & perform various tasks
The main tasks i am focusing on are:

1: Determine all the different obstacles present on the track.

2. Taxi should be able to detect the following obstacles:

*Up-hill; Plain; Down-hill; Bump; 2 consecutive Bumps*

3: Stop Taxi to *take in one passenger only when a Bump* is detected. (It represents that
passenger is standing for a taxi).

4: Stop your car to *drop all the passengers only when 2 consecutive bumps* are detected.(It
represents destination of passenger).

5: The passenger limit for the taxi is 3.

6: Display on LCD the type of terrain on which the taxi is moving...eg. if it is on Bump, LCD must display 'BUMP'.


Help me to with the hardware modules & *main logic*
Any other innovative suggestions are welcomed..

talk with google about their automatic car

Sounds like fun, but why have you taken on a task when you don't, apparently, know how to do most of it?

thanx....well its not for fun...actually its for a competition.

My main aim is to get the idea on how to sense bumps; uphill etc...and more precisely how the BOT will distinguish btween them & display the type of terrain.

Deepesh,

I did a little work with accelerometers in my embedded systems course not too long ago. This will be a much easier project if you take your time and choose a good accelerometer.

The one that I used had one particularly nifty feature: an interrupt output that would send a signal out only when the accelerometer detected a large enough impulse. Think of it like a "shock sensor". It was meant to be used to sense potential damage to the instrument, but, in your case, you could tune it such that it is sensitive enough to sense the bumps. Easy mode.

As to uphill, plain, and downhill detection: Try mounting the accelerometer wherever you want it on your car. Tilt the car to the level you would like to call "uphill" and see what the output looks like. Do the same simulaton for any other terrain you want to sense. Then look at the data, and ask yourself how you can tell the difference between the three cases?

I would look at:
http://am.renesas.com/company_info/carrally/download/index.jsp

especially this:
http://am.renesas.com/media/company_info/carr...load/Program_Explanation_Manual-kit07_38a.pdf

That car is not meant to read "bumps" but rather lines that cross the path of the car. A simple substitution of sensor(s) like an accelerometer or potentiometer or... could also sense bumps.

A fair amount of code is given out and explained as well as sense and control concepts.

You could use shock sensors or you could use reflected light back of the road or you might choose microwave sensors and then run caculations in the backgound as a combination of all three plotting some type of graph or storing this data in ram over a given time period or relative to geographical co ordinates with possible time stats

Looking at this project you would most definatley need constant satelight, rf navigation communications as a separate thread One processor running your sat nav functions Another perhaps running the microwave sensor routine, distance to ground and accelerometer readings

RTOS would definatley spring to mind here as part of your project software "Real time operating system" LOOK UP FREE RTOS as part of your research I trust you c, c plus plus is excellent as should be your knowledge on topics such as fuzzy logic and finite state machines together with interrupts both software and hardware

You will also require extensive knowledge on subjects such as bus communication systems I2C, one wire bus,3 wire bus, serial comms, lcd interfacing, possible pwm theory with code examples and also servo theory, reading addressing and writing to external memory, digital rf transmission and reception

This is very complex project that you are undertaking

Best of luck with this

using accelerometer would be good at primary stage...since i'd not workd on such project b4...

but i'm not able to think abt it...that how the bot will differentiate btwn the diff. terrains.....although its quite easy to detect them

"Its quite easy to detect them"? I'm not sure what you mean exactly... if it's easy to detect them, wherein lies your problem?

Lets look at this from another angle
If you know what the height is from lets say the inner wheel arch to the ball joint of the drivers or passengers wheels on a level plane empty get your answer to this and then you do exactly the same with one passenger then two passengers you have all told three answers
If you then drive along the road and the car hits a bump what happens to the ball joint relative to the wheel arch Answer it decreases does it not

Now you have four answers
So knowing this ,anything below a certain height in retrospect to your wheel arch will initiate a pulse Now you have bump sense within reason
Or if you created a strip marked in black and white which you attached to the shock absorbers vertically such that it would appear as a rectangular strip with wholes spaced evenly vertically and an infra red transmitter, receiver pair of diodes your could then count number of clicks That would be another method without using accelerometer
Might this be a possible answer or solution to your problem or as someone has suggested a potentiometer which measures the resistance

thx Mark..

yeah one f ma frnds also suggested to think about infrared emittr-recievr....it would be easy by measuring resistance continously & feeding it to comparator that would acknowledge with a 'Yes' or 'No' when sensing bumps,etc..

i will work out with them & post the result..