5/14/2008

I got my first gps today, the Garmin eTrex Legend for use hiking and boating. I plan to use it in my kayak. My initial goal is to determine how fast I go in my kayak at my steady pace. I have a Hobie Sport that uses pedals in addition to a paddle; I usually use only the pedals, except for dockings and take-offs.

To start with, I used the gps trip function while I drove to Wal-Mart and compared the gps distance to my car odometer. They were within 1% of each other, and other measurements yielded similar results.

The other day I used Google Maps, with the pedometer function, to estimate my kayak speed. I was able to see my usual dock at Poplar Springs and the spots across Tellico Lake which made up a timing route I ran. The calculations showed about 3.0 mph one way, and 3.9 mph coming back. I suppose the current must be about .5 mph, so it appears from this estimate I am doing about 3.5 mph. I hope the gps confirms this rate, as it seems fast for a novice like me.

I have my usual vision problems with the gps. Without my glasses, which I don’t wear while kayaking or running on the treadmill, I can’t really read the screen. I don’t need to, however, as I can just create waypoints, I think, by simple button presses, and I can turn it off whenever I want and still preserve the trip information. It would be nice to have some audible prompts.

The same vision problem plagues me with the Ipod Nano, so I don’t even use it. Everything is menu driven with several layers, and without audible prompts it is just too difficult to know exactly where I am in the menus. That is why I use the Shuffle instead, as I can just press the buttons to activate its limited functions. For the same reason, I wear a “talking” watch so that I can hear the time and not worry about my eyesight or glare from the sun.

I used one of my IPAQ screen protectors on the Garmin (I just trimmed it after sticking it on), and put the unit in a sleeve with the lanyard. I’m still not sure if I’ll stick in a dry bag or have it looped around my neck when I’m on the water. There are lots of functions in the eTrex, some of which I will probably find useful, but for now I’ll just use the trip calculator. The number of functions is a bit overwhelming, so I’ll slowly ease into them.  [Added on 6/19/2008: I got about 17 hours out of the 2 AA batteries in the eTrex.]

5/15/2008

How I Transport my Kayak

Before I bought my kayak, a Hobie Sport, I had to figure out a way to transport it. My car is small, a Chevy Cavalier, and I needed to be able to load/unload the kayak by myself. It weighs about 48 pounds and is 9.5 feet long. I don’t have the strength to get it on the car roof without some help. I thought about getting a small trailer, but I really didn’t want to drive pulling something along. I also didn’t want to chew up so much of my limited garage space. I eventually decided on the Thule Hullavator, and it has worked quite well. It takes me about 1 minute to unload the kayak, and about 3 to load it (more time to fasten the straps than to unfasten). It took some trial and error to find the best way to run the straps. The two that hold the kayak to the Hullavator are easy, but the third strap that runs the length of the car, front to back, wasn’t obvious to me. My solution is to run a strap through a stern scupper the length of the boat and out the pedal opening while the Hullavator is in the “down” position. I fasten the strap very loosely, and push the buckle up into the pedal hole. After I push the Hullavator into the “up” position and the boat is now on top of the car, I unfasten this third strap, run it around the Hullavator for support, and then refasten it. This method avoids having to climb up on the roof or be 7 feet tall to reach up to do all the work. I have to lift the kayak only about 22 inches off the ground onto the Hullavator.

You have to make sure you have enough clearance in your garage if you use this device, although you could load/unload outside the garage. My garage door clearance is about 84.5 inches; the garage door opener is even lower, but I can put my car in off center and miss it. On my car, the top of the kayak when loaded is about 78.5 inches from the ground. However, it is still a tight squeeze to get the kayak on the roof and avoid the opener, because at one point during the process the kayak is on its side above the car before the Hullavator turns it upside down. Also, you may bump into your car antenna, so watch out.

Here are two photos of the kayak mounted on the Hullavator on my car.

GPS Measures Kayak Speed

5/17/2008

In a recent posting I mentioned that I recently got my first gps, a Garmin eTrex Legend. I took it out in my kayak to measure my steady rate in the boat. I had previously estimated, using Google maps, that I maintained a pace of about 3.5 miles an hour independent of the current (which the day I measured seemed to be about .5 miles an hour). What did the eTrex say? I made 4 runs: out and back, then out and back on a course about 90 degrees from the first course (to eliminate the effect of wind and current). The speeds (in mph) were 3.7 out, 3.2 back, 3.6 out, and 3.1 back. So it looks like my steady pace is about 3.4 mph. Not quite as fast as I would like, but good enough for now.

Working with OCR

So why not a completely different topic? Today I needed to convert some hard copy documents to text. I used a free OCR program, TopOCR, and my cheap scanner. It took a couple of tries before the process was successful. To my surprise, I got better results by using a low dpi rate: I dropped down to 300 dpi before the ocr performed well. In addition, I turned off auto-cropping, even though the preview screen sure looked like auto-cropping should work. At 300 dpi and no auto-cropping, the pages, which were simple printed text, converted to pdf flawlessly.

As part of my project, I needed to convert pdf to MS Word. I tried two of the better known free online services: Zamzar and KoolWire. I only tested simple text pages, and I my experience matched that of most of the online reviewers (look here). Koolwire converted flawlessly and quickly (a couple of minutes), whereas Zamzar put everything inside a frame (easily removed) and took a few hours. I’ll stick with Koolwire.

A Concordance Generator

5/29/2008

A little over a year ago I did a small project to make concordances for some of Joseph Conrad’s novels. I used the electronic ascii texts in the public domain, most of them from Project Gutenberg. These are not authoritative, scholarly versions, but they are suitable for my word counting purposes. I wrote some program snippets to prepare the text for word recognition, and then a program to parse all the words.

There are several decisions about what constitutes a “word” when making a concordance. Here are some examples:

• Hyphens: deep-purple, jet-ski, double-dip, grand-father
• Hyphens at line end
• Numerics: 1^st, 3^rd , 1940’s
• Apostrophes: didn’t, ‘tis, Joe’s
• Spelling variants: color, colour
• Abbreviations: st. for saint

I ended up first adjusting punctuation, and then used UltraEdit to “normalize” all kinds of typographical oddities. For example, Conrad often uses double hyphens, and hyphens at the end of sentences. Although I could have written my own code to handle all these situations, I saw no need to replicate what UltraEdit could already do quite easily. By the way, I’ve used UltraEdit almost since its first release (1996), and have found it to be indispensable. After processing with UltraEdit, I now have plain ascii text where a “word” is always terminated by a space (decimal 32). My program then finds all the words, and writes out a CSV (comma separated values) file with the word, its absolute position in the file, and the length of the word, with one “word” per line. It takes less than 2 seconds to process a Conrad novel this way. The final step is to load this CSV file into an SQL database (I use postgres). I add a few identifiers to the sql record (an id to tell me source book, and anything else I like).

It is easy to use the database to do word counts, sorts, and lots of other things (like “neighbor” words). I also wrote a simple “context” display to show the x number of preceding and following words for any user selected word. Certainly a Google search can do some of the same things, but it isn’t geared to do word counts.

I have downloaded a few of the free concordance tools from the internet, and they work well enough, but none of them adjust for the issues mentioned above.  I like to have more control over the process, and using my own code allows me to collect whatever attributes I like and load them into a database.

As a by-product of my work on a concordance maker, I also had to produce a concordance for MS-Word documents. I was surprised not to find any free software on the internet to produce automated word lists from Word, so I simply used my own. I saved a Word doc to plain text and inserted line breaks (a Word option). I did discover that Word converts a few characters (like apostrophe—Word uses decimal 146 instead of 39, at least in the documents I was using) to unexpected values.

One Result from Concordance of Works of Joseph Conrad

So, what if anything did I find after building a concordance of 9 of the novels of Joseph Conrad? I used the following books:

• Victory
• Lord Jim
• Heart of Darkness
• Almayer’s Folly
• Chance
• Nostromo
• Outcast of the Islands
• Secret Sharer
• Shadow-Line

Here is one thing I observed and is presented without any interpretation.

I was looking at a few words which rarely appear in Heart of Darkness, but are not so uncommon in the other stories. The words “perfect”, “strong”, “visible”, and “dare” (coupled with “dares”) appear a total of 7 times in HD. After adjusting for the difference in total word count among all the books, I would expect these words to appear about 149 times in the other 8 books. However, they appear 515 times, or almost 3.5 times what we would anticipate if we extrapolated from HD. If I look at the books one at a time and compare these word frequencies to HD, only Secret Sharer is similar to it. There are probably a few other words that fit this pattern (ignoring words obviously peculiar to HD, like proper nouns and “Intended.”)

There are some variants of these words (“perfectly”, “imperfectly”) that should be included, as well as an examination of negation (“not perfect”).

Paddling and Peddling

6/26/2008

Maybe I should have said, “Paddling While Peddling.”  I’ve been experimenting paddling while peddling, and have found that paddling increases my steady pace, at 84 bpm,  from 3.4 mph to about 3.7 mph.  What I call my “burst” speed, which I so far have sustained for 32 minutes (and could do more, I’m sure), is just about 4.0 mph.  Today I did 65 minutes and averaged 3.7 mph; I paddled on and off for the whole trip, resting for about 60-90 seconds every so often.  I also used paddling gloves for the first time, but I’m not convinced yet they work as well as just wrapping my thumb in gauze and tape.  My new touring paddle does take less effort than my old paddle; on the other hand, I think I can actually go faster with the older one.  By the way, my steady peddling rate of 84 bpm is exactly half of my pace on the treadmill.  A side benefit of the paddling is that my calorie burn is at least double what I get just peddling.

The ST Turbo Fins

July 30, 2008

I put the ST Turbo Fins on my Hobie Sport and I was very disappointed with what happened.  Yes, the fins added greatly to my speed: my cruising speed went from around 3.4 mph to slightly over 4.0 mph.  I used my gps and a stopwatch to get an accurate reading, and I went twice back and forth to eliminate current and wind effects.  Both runs yielded almost exactly the same times.  So why was I so disappointed that I removed the fins and replaced them with the originals?  After all, I attained nearly a 20% speed increase at close to the same number of “beats” per minutes (the fins are advertised to yield a 10% speed increase).  The speed increase of course requires more resistance when peddling, but the increase was so great that I could not sustain the effort for more than 30 minutes at a time.   With the original equipment fins my back gets sore before I get tired or my heart rate elevates.  The resistance with the turbo fins is so great that I could not paddle effectively while peddling.

GPS PhotoTagging

[Note: since I wrote this, I have abandoned the use of the Merax.  I now use some of the user supplied Google Maps tools to freehand my route.  I use the Garmin Etrex to provide start and end points.  The Merax is unsupported, the code it generates is cumbersome, and device itself sometimes "resets" itself during logging.]

I saw some examples of tagging photos with gps info and displaying the images on Goggle Maps, and I thought I’d like to try it on  some of my pictures.  As I’ve mentioned above, I do have a portable GPS unit (Etrex), and I could write some code to take its log files and synch with the digital photos, but that seemed like a lot of work.  Instead, I took the easy way and purchased the Merax Photofinder RAW software and mini-gps unit.  The unit does nothing but create a log file; there is no information visible in the unit except the on/off lights.  Everything needs to be transferred via usb to the PC.  The unit itself is small and easily fits in a pocket.  The software, although lacking any useful help documentation, is very easy to use and does quite a bit.

You start by synching the gps clock and your camera clock with your pc.  You leave the gps unit on during the trip (or anytime you want to record position into the log), and when you are done you upload the log and your photos to your pc.  The Merax software then matches the time stamps (I have no idea of how it resolves “close”), then uses Goggle to attempt to find a place description.  This information is then added to the photo EXIF (if you so desire).  Finally, and this is the best part, the software supports an easy export to your pc that can be uploaded to your own internet site.  It does this by supplying a Google Map API key (you specify the URL of the upload site).

It is necessary to ftp the exported files into exactly the same directory structure on the internet site.  That’s about all it takes, and you have a Google Maps application showing the points and photos of your trip.

The Merax software lets you supply descriptions and has one primitive method to try to find a location.  You may find useful an EXIF editor (there are some free ones on the internet) to adjust the lat/long from time to time.  It also supports Google Earth, but I haven’t tried that.

The only bug I’ve found so far is that the “breadcrumb” path, which shows on your PC, does not appear in the files that get exported (although the software claims to support this function).    A friend of mine with expertise in these matters gave me two lines of code to fix this problem.  Thanks, Jim.

All in all, this is an incredibly easy way to put your digital photos onto a Google Maps viewer, and to place the application onto your web site.    However, although the software can be made to work on other gps log files, the files would have to be manipulated to be in the required format.  None of this is documented.  So the package of unit+software seems to me to be the simple way to go.

Although I know a lot about Windows programming, I haven’t done any web applications.  However, there is so much material easily accessible on the internet, I found it not too difficult to get started with Google Maps. I used the Merax software as a launching pad, and then used a community add-in to get the lat/long for freehand routes, and for single points.  I tweaked the place markers to be “draggable” to circumvent the problem when the markers are so close together you can’t click on all of them.  I used some other free code to add a label to the map.  You can see the results at my web page.

More on the Thule Hullavator

11/5/2008

My Thule Hullavator has been just what I wanted: a way to load, transport, and unload my kayak without any help or a SUV/truck.  However, after about 200 loads and 200 unloads, one of the “arms” got out of alignment and would not stay in the “lock” position.  I had to tie it down (a 30 second operation), lif the kayak onto it, untie the rope, and then (as usual) raise the Hullavator to load the kayak onto the roof rack.  The post on which the arm rests also got a little of alignment, and made it a little difficult to get the locking pin in and out.  The device was still useful, but it was not performing as it should.

I contacted Thule, and they were very helpful.  They sent me a completely new Hullavator, after I sent them proof (a picture of the post sawed in two) that I would not have two working Hullavators.  However, the new one was defective, right out of the box, with exactly the same misalignment problem so that the arm would not stay locked.  I had an authorized Thule dealer do the installation, and he verified that the device was flawed.  I notified Thule of the difficulty, but got no reply.  Good thing I still have the rope.

4/30/2009

I again contacted Thule, explained the problem, and after providing them with photo documentation (via Picasa), they sent me another new Hullavator.  I assembled it, made a couple of small adjustments, and mounted it on the rack.  I then tested it both with and without my kayak; everything worked flawlessly.  My original Hullavator gave me about 200 loads and 200 unloads (one season), and I can only hope that this one lasts longer.

8/7/2009

Hero Surf/Helmet Camera & Other Cameras

I bought a GoPro Hero surf/helmet waterproof digital camera recently, and today tried it out on both of my kayaks.  The suction cup mount worked well on my Epic GPX, but wouldn’t take on the Hobie.  The Hobie just doesn’t seem to have any truly flat spots to create a solid seal on the suction cup.  I did manage to cram the mount and camera into one of the Hobie’s fishing rod holes, although it wasn’t terribly stable.  So then I got some Velcro and built a flat platform at the front of the boat.  I may need to modify it so that I can orient the camera in any direction, and perhaps make the platform thicker so that the camera sits higher.   In both boats I pointed the camera off to the side in the hopes of photographing the shoreline rather than just water and sky.  In any event, the pictures were better than I hoped for, but  significantly worse than even my cheap digital camera.  The pitch and roll of the kayak, along with some wobble from the suction cup mount, cause lots of the photos to be slightly out of focus.  Of course with photos being taken every 5 seconds, all I really want is a few decent shots; the rest go into the trash bin.  An hour paddle can yield 720 pics at the 5 second rate,  so with just a little luck there may be some gold in that pile.  The Hero does provide a record of the trip, and I suppose that is its prime use, and it is nice not to have to wrestle with a camera in my conventional kayak.  The Hobie is so stable that it is a simple matter to take pictures with a decent digital camera.

I’ve also used two inexpensive transparent waterproof “camera bags,” but I am not satisfied with them.  Although they are waterproof,  you really need to leave the camera turned on because the moving of the lens for zooming or on/off tends to unsettle the camera in the bag and leave it an unusable position.  They do work, just not that well.  On the other hand, I’ve had good luck with the Kodak PlaySport waterproof video cam.   It has undergone a dunking with no ill effects (good thing I had a tether on it), and its HD video looks good.  Of course you can’t focus it, it has no viewfinder, and the lcd screen is impossible to see in the sun, but if you just point and hope, you can get a decent video.  It picks up sound surprisingly well.  I’ve also used a waterproof Pentax Optio for a still camera, and its pictures are acceptable.  It is difficult to focus (lcd screen in bright sun) and manipulate with paddle gloves and sunglasses, but it can handle the kayak motion without blur most of the time.  Sometimes I use “burst mode” to avoid motion blur.

Another Kayak

I bought a second, conventional recreational kayak: a 31 pound EPIC GPX composite.  It is light enough that I can carry it, lift it to my roof rack, and do everything I need to do by myself.  I added some standard inexpensive carriers on the roof rack, on the side opposite the Hullavator.  The car can now carry two kayaks at once, one in the saddles and one in the Hullavator.

A Great Monte Carlo Simulator

I’ve used a resampling software, rstats, from Julian Simon and Peter Bruce, for several years.  It is very easy to learn and quite powerful.  However, it is so old it is no longer supported.  Happily, there is a giftware terrific enhanced rewrite in Java called “statistics101.”  The website for it,  www.statistics101.net,  has tutorials and a useful forum.  If you are interested in the subject, visit the site and download the program.

Here is a little example using statistics101.  I like to play Mahjong Titans using the “Turtle” layout of 144 tiles.  The very top tile sits on a pile whose next layer has 4 tiles.  I noticed that the layout doesn’t seem to be random, because almost every game has a tile in this layer that matches (and so is covered by) the top tile.  So how often should that happen?  It is easy to do a direct calculation.  Let any tile be the top tile, which leaves 143 other tiles.  Out of these we pick 4 and see how often none of the 4 will match our top tile.  There are 143C4 ways to chose these 4, and 140C4 will not have a match to the top tile.  So the ratio of 140C4/143C4 will be the probability that the top tile will not match any of the 4 directly under it.  Another way to think about it is to say, ok, I can pick any of 140 tiles that don’t match for my first choice (out of 143), 139 (out of 142) for my second choice, 138 (out of 141) for my third choice, and 137 (out of 140) for my 4th choice.  This yields the same answer: 139*138*137/(143*142*141)=.918 that the top tile will not match any of the underlying 4 tiles.

Let’s look at how to simulate this process in statistics101.  Here is a possible way:

copy 0 nomatchcount    ’initialize nomatchcount to zero

urn 3#0 140#1 A    ’0 for the 3  tiles that match the top, 1 for all others

repeat 100000

shuffle A shuffledA   ’randomly order the contents (143 tiles) of A and put them in    shuffledA

take shuffledA 1,4 B      ’take the first 4 tiles from A and put them in B

count B=0 anymatch ‘count how many 0 which are the possible 3 matching tiles and put   the answer in B

IF anymatch=0

add 1 nomatchcount nomatchcount ‘we did not get a match, add 1 to the counter

END  ‘of the IF command

END  ‘of the repeat

DIVIDE nomatchcount 100000 estimatedprob ‘how often we got nomatch divided by our 100000 trials

PRINT estimatedprob

The first 3 runs of this program yielded .9169,.9196, and .9180 for our estimated probability.  That’s close to our directly computed answer of .918.  In my experience, the Turtle layout has the top tile directly covering a match at least 80% of the time, which means that the layout isn’t randomly generated.

Garmin Forerunner & Kayaking

I’ve added a Garmin Forerunner to my collection of gadgets, and find it very useful while kayaking.  Not for tracking and mapping my route, but for measuring distance, speed, and heart rate when I go for an exercise paddle.  It tells me I usually average around 4.0 miles per hour (in my 13 foot EPIC GPX) for a 3-4 mile outing, with an average heart rate of around 104 beats per minute.  The “training center” software included with the Forerunner is nice, as it lets you see a map of your route, totals for various categories, and allows user definition of 10 “zones” for speed/distance and heart rate.  Below is part of its display.