# How to find deviation of vehicle trajectory points from median?

1511
10
05-20-2018 12:33 AM
New Contributor

Hello,

I am working on a project where I need to find the deviation of the vehicle trajectory path from centre line of median of the road. I have a large number of coordinates collected using GPS for both the trajectory path and the centre line of median. Is there any tool or some other way to have the deviation values in meters in the form of  a table?

Regards

1 Solution

Accepted Solutions
MVP Legendary Contributor

Alice.... long story... have a look at the data and I can explain in detail but

• fixed the data to get long/lat in the right order
• trimmed to the common extent
• projected to utm 46N wgs84
• got the x, y values in both tables
• converted the median points to a median polyline representing 1 'Feature'
• used near on the vehicle data to get the near location distance and direction to the line

This is the southern end

The data are attached in an xls for your perusal

10 Replies
MVP Legendary Contributor

If you want a mean deviation in planar units, you will have to project all the coordinates to a suitable projection (eg. UTM).  Once you have that in tabular form, you calculate the distance in another field using Pythagorean theorem.  Those fields can then be use to obtain simple statistics using builtin tools.

I am not sure why you have ArcGIS Online as the chosen place.  Perhaps you meant to post in Geoprocessing‌ or ArcGIS Pro

If you were using ArcGIS PRO and not ArcGIS Online, then you could use

Table Tools.... or

New Contributor

Thankyou so much. I wasn't sure which place would be apt to put up the query. I am using ArcMap and quite new to this. Can the median be converted to a smooth curve and shortest distance from each point of the vehicle trajectory be calculated from the curve? I have tried using the near tool for this but it takes only point values.

MVP Legendary Contributor

The last time I worked on this  stuff we converted the centerline to points to facilitate the equivalent of today's Near tool.

If the centerline is fairly detailed representing the true shape (particular in the area of circular curves) then you won't have much problem.

Once the gps are associated with a near position, fine-tuning can begin.

Perhaps if you have a sample of the road segment and its gps denoted vehicle trajectory data, it would help.  A picture for now and/or a small sample in a geodatabase

would be interesting.

New Contributor

I have attached a .xlsx file along with my question. Kindly see if it helps.

MVP Legendary Contributor

Alice... working on it.  What projected coordinate system do you use for your area?

I assume current values represent GCS WGS84 and longitudes and latitudes are both positive as shown.

There is also incomplete overlap between the two sets which I presume is correct

New Contributor

Projected coordinate system is WGS84 46N. Yup, lat and long values are positive.

MVP Legendary Contributor

Your X and Y's are switched  ie Y_rd_median isn't the Y values which should be longitude.  Longitude is X

That slowed me.  I have stuff projected now and will try to dissect the overlapping area.

I suggest you do the same so you have the common extent of both since they don't overlap directly in the north/south direction

MVP Legendary Contributor

Alice.... long story... have a look at the data and I can explain in detail but

• fixed the data to get long/lat in the right order
• trimmed to the common extent
• projected to utm 46N wgs84
• got the x, y values in both tables
• converted the median points to a median polyline representing 1 'Feature'
• used near on the vehicle data to get the near location distance and direction to the line

This is the southern end

The data are attached in an xls for your perusal

New Contributor

Thank you so much for the help! Exactly what I needed.