Raj,

Let's assume you have two stepper motors, one per axis. One knows how many steps each motor makes per revolution (N_x, N_y). One also (hopefully) knows the linear distance that is equivalent to one rotation, for each axis (L_x, L_y)

There's no magic, this is basic plane geometry.

If your initial point is (x_0,y_0), for a number of steps on each motor (n_x, n_y) your new position is: (x_new, y_new) where

x_new = x_0 + n_x*(L_x / N_x)

and

y_new = y_0 + n_y*(L_y / N_y)

the angle, theta, between your old and new points, using the x+ direction as the zero heading, and measuring in a counter-clockwise sense is

arc_theta = n_y*(L_y / N_y) / n_x*(L_x / N_x)

You may want to investigate more sophisticated heading measurement to avoid pesky singularities and ambiguities.

LabVIEW can manage all of the above trivially.

I am using same thing, but i am use stepper motor count to find angle instead of revolution it will give me x,y coordinate in only forward direction when some time it will give right angle but some time it gave wrong angle

can you give me best way to find angle and coordinate

Dear Raj

the topic you are looking for is called:

Odometry and Dead Reckoning.

There are plenty of web pages dealing with this problem.

Have a look at the web page below, might be helpfull:

http://www.seattlerobotics.org/encoder/200010/dead_reckoning_article.html

In general, odometry by counting/measuring the revolutions of your wheels is not a good method to determine your pose: position and orientation.

Unfortunately even small mistakes generate a large deviation after driving even short time.

Regards

Nils Goerke

Raj,

I confess, I am not at all clear now what your system looks like. Nils is perfectly right - wheel-slip, encoder errors, and so on all conspire to make odometry the worst (?) method for deducing position.

Do you have any other options (GPS, etc.?) And how accurate do you need the position to be?

Do you have a link to the system's description?

(Is this a vehicle with driven wheels? Or an X/Y plotter bed? Or something else? And if your stepper motors provide angles (!) then in the LabVIEW something must be reading the absolute angle (gray codes?) or doing some counting.)

More details please.

May be, you both are right

i am using stepper motor whose angle is 3.75 so when 92 pulse is given to stepper motor it will rotate 1 round

i am measure pulse count using controller then it will transmit measure data through zigbee, another zigbee attach received data and then whole calculation is done in LABVIEW

using the same pulse i am calculate traveled distance by My small robot it show perfect reading.

but same time it will gave me wrong angle so can you suggest me how to calculate angle

Raj,

And may I presume two wheels? One on the left, one on the right, of the robot?

So, when driving the wheels with an equal number of pulses, you'll get identical values for the distance travelled by each wheel.

So that matches with your description of, "using the same pulse i am calculate traveled distance by My small robot it show perfect reading."

What you want, I think, is to know the heading vector at any time (or the angular displacement from an origin) when both wheels have been driven with an arbitrary number of pulses.

This is quite hard.

You will have to track, at a finely divided timescale, the number of pulses that each wheel has logged (or, conversely, the angle through which each wheel has turned).

If you are driving each wheel with different pulse rates, then your craft will execute curved paths. You will need to find the partial arcs that describe the trajectory of the robot between manoeuvres and then add them vectorially.

This is not simple.

If you are driving the wheels such that you either have rotations *or* translations, then it's much easier.

But I suspect that ResearchGate isn't the best medium for this - would you not be able to find a willing soul at the University that you attend?

Yes, you are right on wheel is in left side and another wheel is in right side.

i have already tried both heading angle as well as angular position from origin. and tried with distance travelled in one pulse. but i have same error

+/- 4-5 cm error is ok for me

Dear Raj

i see four possibilities for you:

1: a "quick-hack" solution:

implement the formulas from the web pages you allready have or from one

of the two pages directly below.

This will be sufficient if you only want a rough pose estimate.

http://www.comp.lancs.ac.uk/~kristof/research/notes/deadreck/deadreck.html#_Toc454510239

http://rossum.sourceforge.net/papers/DiffSteer/

Of course, you will have to adjust the parameters to the geometry of your robot.

2. You need a more sophisticated solution:

Then, unfortunately you will have to go through the theoty behind, to understand

how the formulas are derived. This will still lead to same formulas as solution 1,

But you have a more precise possibility to adjust the respective parameters,

and thus the residual error will be smaller.

This is a fine way to implement the dead reckoning for a short distance

pose estimation.

Google with the key words:

localization where am i odometry siegwart

and take the first link to a pdf file, and there section 5.2.4

(The complete link is toooo long to put here)

3. You really, really need an improved dead reckoning.

Then, you will have to go really deep into theory of the kinematics of your robot

including wheel slip, surface structure, encoder errors, time skew between

the motors, pose of the wheels (are they really parallel?), structure, size, and

postion of the rest of the robot (does it has a castor-wheel?), coulomb friction

while driving, ...

There are research groups dealing with all these aspects, and generating a

cure for most of these problems.

They will require that you read the respective publications.

4. Chose to take a complete different method for estimationg your position.

Thanks, may be it will helpful to me

Odometry and dead reckoning is important for sure, to estimate the XY position using the angle measurement of the wheels: they are fast and they will work at any condition; as long as the wheels are turning (i mean the environmental issues, disturbances, etc.) However, they DO NOT detect the slip of the wheels or ARE NOT quite accurate at the task space.

So, to improve the XY positioning of mobile robots, most of the time a global measuring unit is introduced: depending on the work space of your device, and the working environment, many applications you can find in the literature I suppose (GPS, Kinect, optical sensors, cameras etc.) To integrate these new sensors with the odometry, the most common technique is to use Kalman filter based sensor fusion techniques. You can search for this technique and I am sure that you will find many articles, publications or applications..

Good luck!

it not possible to use GPS and kinect because my mobile robot size is 12 cm x 12 cm x 10 cm (length x width x height ) and it has to pass through the closed bounded area like 25cm x 25cm (width x height) or 30 cm round shape area

in this, i am count the pulse of stepper motor and this measured pulse is transmit through zigbee to user end (10m away) and whole calculation i am done in LabVIEW software

For small sized robots, I have seen some examples where they obtained good accuracy with optical flow sensors. You can check the following paper.

https://www.researchgate.net/publication/221072760_Robust_Mobile_Robot_Localization_Using_Optical_Flow_Sensors_and_Encoders

Moreover, gyroscope data might be also useful for you, maybe.

https://www.researchgate.net/publication/2400799_Experimental_Evaluation_of_a_Fiber_Optics_Gyroscope_for_Improving_Dead-reckoning_Accuracy_in_Mobile_Robots

Last but not least, if you're working in a bounded area, you can use laser pointers or devices like that to estimate your position depending on how far your device is from the specified boundaries maybe.

Your calculations about this step motor sounds right, but just not accurate enough. You just need to verify these calculations with a global measurement.

Conference Paper Robust Mobile Robot Localization Using Optical Flow Sensors ...

Article Experimental Evaluation of a Fiber Optics Gyroscope for Impr...

Dear Raj

if your robot is too small for a kinect,

you can use a small laser scanner

e.g. Hokuyo

http://www.hokuyo-aut.jp/02sensor/07scanner/urg_04lx_ug01.html

and take a classic localisation method,

e.g. amcl (Adaptive monte carlo localisation)

Then, you can use a SLAM mapping algorithm as well,

(SLAM Simultaneous Localisation And Mapping)

see. g-mapping for software and theory.

I wish you success

Is this possible to measure angle of rotation with help of stepper motor pulse count

because there no space to mount any other sensor like odometer, kinect, gyroscope etc.

I am using two stepper motor to drive small robot

base on the stepper motor pulse count i take difference between this two pulse to measure direction of angle of rotation (left, right , straight )

is it true method to find angle of direction

Hey Raj, I am doing a similar project to find the new X and Y coordinates using Labview and the Laser Tracker. Is it possible that I could take help from you??

ya 

you can use gyroscope sensor so you can find roll,pitch and yaw angle based on that you can measure X,Y co-ordinate 

well im trying a new approach, with laser tracker I can find the position and orientation of the robot, but my problem is that I'm new to LabView and the coding looks very difficult