Sounds risky, but clever short term:


Talk about being disorganized - 
Sometimes I think to post here on Google+

Sometimes I post on Blogger http://alanmcdonley.blogspot.com/2015/07/memories-of-limping-robots.html

Sometimes on forums like raspberrypi.org, mikronauts.com, or dpreview.com

Photos usually go to Flickr:  https://www.flickr.com/photos/85977298@N07/

Yesterday I started playing around with wix  and tried to answer the question "Who Am I / Introduce Yourself": http://alanmcdonley.wix.com/curious#!about/cbga

I used to put everything on geocities till Yahoo bought that free site and shut it down.  Then I used the free site koolhost, but they "lost" my website after five to seven years.  I've been "homeless" for a few years now, and I can't seem to settle in one place now that I have invested in a few spots.  

Alan

My head has been spinning robot thoughts lately.  

Solar Powered Raspberry PI B+ Thoughts:

AA cell can store 4700 Joules or 1.3 watt-hours or 1083 maH at 1.2v

2500 maH at 7.2v (6 NiMH cells) = 18 watt-hours ~= 19000 Joules

Shaded south facing window sill position provides 1.43 J/cm2/day 
  (Columbia University, Electrical Engineering 
   Technical Report #2010-07-15, July 2010 )

voltaic systems V15 USB battery 
   4000mAh, 15 Watt Hours, 5V/1A USB

voltaic systems 6W Solar Panel (at 100mW/cm2 irradiance - 0.1J/cm2/second):
   8.7x6.9” 8.5oz  (385cm2) 
   About four hours in bright sun will charge the V15 battery

  
So 385cm2 panel at 1.43 J/cm2 indoor = 
  = 550 Joules/day indoor shaded south facing window
           = 152 mwH 
  = 25maH at 6v indoor per day
    
Raspberry Pi B+ min draw of around 200-250 ma and perhaps 1A other times.
 
If robot requires 250ma min and 1A with WiFi or Camera at 6v, 
    need 10 indoor days to get 1 hour of very low power execution,
    and 40 days to get 1 hour of full execution
    or can power robot just 6 minutes per day at minimum power configuration.

Four hours outside in direct sun is a better story (but still lacking):
    will allow 16 hours at minimum draw.

There is a board called SleepyPi with a RTC that allows setting an alarm time, putting the Pi to sleep, and it will wake Pi up at the alarm time.  If the bot sleeps from 8pm until 6 am (10 hours), and runs at high power draw for three minutes an hour to send/receive “its” mail, or to drive to the sunny south facing window, four hours in bright sun might be able to stretch to cover a whole day in the life of a Raspberry Pi B+

#SolarRobotics   #Solar_Raspberry_Pi   #Indoor_Solar_Robot_not_likely  

If you find you cannot see a Facebook "Say Thanks" video someone shared with you, (just sits there loading), from your desktop browser(s), even after cleaning cookies/history and restarting/rebooting... try watching it on your mobile device.  You may get to see it (once, but it may do the same loading behavior the second try).   

I'm not even on Facebook but get lots of Facebook support calls...

Test Results Posted:
Raspberry Pi 3 Model B compared with Raspberry Pi B+ 512MB
Using CMU PocketSphinx Speech Recognition:
https://goo.gl/RrGgCm

The Video (2m 30s) is posted: https://vimeo.com/169445418

Pi 3 is going to be a real game changer for my robot.

Audubon Society Complicit In Torture - One Million Birds Each Year

Recently Fairchild Tropical Gardens (south Florida) held a bird festival with birding tours "led by Tropical Audubon Society experts”.  One was billed as a Half-Day tour to “include a visit to Cape Florida’s bird banding station”, “on this walk through restored coastal habitats”  of Bill Baggs Cape Florida State Park.

It was on this tour that I discovered I am vehemently opposed to bird banding or bird ringing as it is called outside the U.S. I learned that this one south Florida station alone has tortured over 27,000 birds in the prior ten years “to help migratory birds.”  

The "help" begins with a period of extreme stress, with possible wing or leg damage, from being caught in a very fine net.  The bird’s struggle to get free uses some of the precious energy reserves the bird has been working hard to build up.  After not a short time, a volunteer bird sadist checks one of three areas of five to nine nets.  On discovering a bird in a net, the human extracts the bird from the net and puts it in a small cloth bag closed with a draw-string.  After all the nets have been checked, the bird sadist returns to the banding station and hangs the bagged bird on a rope line strung above a work table.  The position on the line corresponds to the number of the net that ensnared the bird.

Eventually a worker will extract the bird from the bag, and inspect every inch of the tiny bird.  The worker puts a band around the bird’s leg, records various size measurements, and the amount of fat reserve the bird has.  Next the bird is dropped, beak down, into a section of pipe to measure the bird’s weight.  After saving the bird from the pipe, it is returned to the drawstring bag, and hung again on the line to await its return to freedom in the area from which it began this nightmare.

The banding station workers enthusiastically told of one bird being recovered daily for over a week, allowing them to determine that this bird had added ten percent to its mere 100 gram (3.5oz) stature before (presumably) heading over the ocean to distant lands.  Another story related that one bird had been “recovered” each year for eight years.  To the question about this station's recoveries, we heard that very few are ever seen again - “much less than one percent”.  

(After more than an hour there, we asked the tour guide about the rest of the tour, and were told “this is the tour”.  We left at that point.)

When I got home, I looked for more information about banding in the US and found:
1) The US Geologic Survey runs a national bird banding laboratory.
2) In the US, over 1 million birds are subjected to this torturous banding every year
3) Various research papers estimate that one half to one percent of the birds banded each year die from the stress or injuries of the process - (50,000-100,000 birds).
4) Between fifty-five percent (uninjured) and sixty percent of (injured) banded birds may get caught by the same banding station a second time within one month of banding.
4) The Audubon Society actively supports banding, collection, and sampling efforts with paid internships, training, and management.

While national research surveys estimate less than one percent lifetime return capture, one paper reported recapture and return capture rates as high as 35% (greater than one year) and 15% (greater than two years).    

I could not find any quantitative measures of banding benefit to birds, or regulation influence.  Typical statements of benefit suggest banding helps the birds by identifying staging and foraging areas.  The most prominent statement of benefit of banding to birds is the US Geologic Survey Bird Banding Lab’s front page attribution that banding helps the development of hunting regulations.

Just days before the tour, I had read in a Cornell University birding publication of the success of tracking migratory birds with weather radar.  I have frequently used the ebird.org site to learn what birds knowledgable folks were seeing at the hotspots I visit.   The quality of population estimation by observation has been studied and can approach capture/sampling methods.  I could not find studies that contrast the benefit to birds, of bird capture over harmless observation counts.  

These days, money has much more influence in politics than scientific data.  It is difficult for politicians to evaluate the bird science lobbyists place before them, but easy to evaluate the amount of lobbying for and against regulations.  In this respect, the Audubon Society’s fund raising to pay lobbyists may well be the actual benefit to birds, and not the Society’s support for the sacrifice of thousands of birds. 

I will not support the continuation of banding, regardless of the esoteric knowledge we may gain, or the touted political power of banding data.  I cannot support the Audubon Society now that I know their complicity in torturing a million birds a year.  I regret my participation in this Fairchild Tropical Gardens Bird Festival tour, and “wish I didn’t know now what I didn’t know then.”
 
Disclosure: I contribute to the torture and death of over 50 chickens each year, and I eat their unborn.  I am a terrible, conflicted eater.  

My head has been spinning robot thoughts again.  

I woke up my RugWarriorPro robot after several years of deep sleep, and refreshed my memory of the specific capabilities and limitations of my 15 year old robot.  

Also revived were memories of the pleasures of having a community of enthusiasts using the same bot platform, to share accomplishments, code, and ideas with.  The RugWarrior and RugWarriorPro community was active for about six years (1997 -2002) with some 600 persons total.  

Student participation in the community was usually through use of a university owned RugWarrior(Pro) bot for a single semester, or through a high school robotics club’s annual contest entry.  Hobbyist entry into the community ranged from purchasing the pc board and gathering all the parts, to purchasing a $600 “brains and brawn” kit, assembling everything into a working robot, and then expanding either the common robot’s brain capabilities through writing software, or adding sensors and effectors with writing the associated interface software.  

There was a 450 page printed “bible” to explain every aspect of the robot from requirements analysis, design, hardware, software, to ideas for further study.  There were businesses selling partial and full kits, along with ready to mate parts for the students and hobbyists. 

After the demise of the RugWarriorPro community, I continued to hope for the emergence of a new robot collaboration of educators, students, business, hobbyists, and social media.     

In  2000, one of the educators from MIT that created the RugWarrior formed iRobot, and soon released the robotic Roomba vacuum.  Hobbyists discovered that the Roomba had a hackable interface port reminiscent of the RugWarrior’s serial communications port.

About that time the tiny, Linux based GumStix computer began selling at around $100.  When iRobot released the Roomba Discovery kit, I was thinking that a GumStix/Roomba combination might birth a new common robot platform and community.  The high entrance cost of using the Roomba as the mobile base may have been a barrier to university and high school adoption.  

Also around 2005, the Arduino processor emerged.  It continues to have a strong following.  The Arduino is inexpensive and well supported as a basic controller and sensor interface, but it lacks the processing power needed to rise above a very basic platform.  An Arduino starter kit, which sells for $50 is 16 times faster than the $300 RugWarriorPro brains kit of 1999, but still has a pea brain amount of working memory (32K bytes to 512K bytes - kilo... ).

I thought perhaps a GumStix/Arduino lashup would catch on.  This pair of boards with sensors, and hacked servos from radio controlled vehicles can create a powerful platform for around $200.  Embedded Linux had a fairly steep learning curve in those days, which may have hindered the GumStix adoption.

The year 2012 may prove to have been the birth of a new era for personal and embedded computing.   The Raspberry Pi (Pi B+ currently selling for $35) is showing very strong among educators and businesses.  The Pi can be used, as is, for learning about operating systems, computer languages, simple digital sensor interfacing, even client-server web programming, and using cloud services (Google, Amazon, and social media).  

It is slow by comparison to even the least expensive laptop, and by the time you add the prices of a power supply, keyboard, and mouse, (perhaps using a students existing television as a monitor), the general purpose educational platform starts to tarnish.  In a school setting, where each Pi “station” would also need a display, the Pi would not seem to be the best choice for general computing education.  

It is in embedded applications such as robotics, that the Pi has the best price-performance showing.  The Pi comes with high speed buss interfaces built-in, that ease pairing with the Arduino for sensor interface and motor control, relieving the Pi to “worry” about higher level computing with gigabytes of memory and a processor which is 50 times faster than the Arduino boards (700MHz-1GHz turbo vs 16MHz).

Educators can use the Pi alone, or an Arduino alone, to teach basic interface and control concepts.  They can then introduce buss interface concepts by pairing an Arduino and a Pi.  The Pi easily supports very high level concepts such as vision, speech, reasoning, and learning.  The price of this low-level and high-level computing “station” is lower than a single laptop, and adds physical world computing to the educational possibilities.  

While competition among businesses can bring down costs and encourage diversification of available features, the diversity also splinters communities. There are many “hats” and “shields” for the Raspberry PI, but most are an immediate interface connector  endpoint, or they try to be an exclusive expansion path, rather than providing true plug-and-play, mix-and-match features in a stack of community standardized interchangeable modules.

Additionally, the Raspberry Pi made a significant change to the general purpose interface connector, which has set back businesses attempting to support Pi adoption.   

It may be that the Internet age both opened embedded computing to the masses, but delayed the adoption of standards which create critical-mass communities of educators, businesses, and hobbyists.  

All this potential is in great contrast to my RugWarriorPro bot.  My bot has not “grown” with age, but my desires for personal bot capabilities have.  What I want in a robot is:
 - Reliably return to a recharging station without assistance.
 - Measures its energy store, and manages energy usage.
 - Capable of surviving extended, complete power outage
 - “On board” programming environment for all components
 - Wireless network connectivity 
 - Mobile platform handles uneven floors and floor/rug transitions.
 - Mobile platform that is “small dog” size
 - Mobile platform can move at some speed “quietly”.
 - Builds a functional model of rooms, doorways, and danger areas.
 - Maintain rough awareness of location (what room, body widths from nearby obstacles).
 - Maintain rough awareness of orientation.
 - Navigates to a goal room or position.
 - Pleasant and understandable text-to-speech capability
 - Strong audio output through a small speaker(s)
 - Directional, far-field, noise canceling microphone
 - Hot-word triggered, word spotting speech recognition
 - GPU assisted vision processing
 - Vision processing - trainable object recognition
 - Vision processing - trainable facial recognition
 - Ability to search topics on internet
 - Self managed local knowledge base of “useful to self” concepts
   * “significant” self historical events
  * rolling short term self action log
  * rolling short term sensed event log
  * add knowledge from learning sources matching a learning goal
  * evaluate value of new knowledge
  * reevaluate value of old knowledge based on weighted criteria
  * reevaluate knowledge value criteria based on status and goal set
  * use human interaction to validate knowledge
  * maintain models for individual humans (mood, uses, tastes)
 - Moderate “presence” to perceived human models at moment and over time
 - self contained chatbot system for human interaction 

Enough dreaming for one post.

Alan

So cool how it flies off and returns.  I am always trying to follow then, and have never thought to wait for them to return to my FOV.

I started running 18 months ago at age 60 with the GetRunning C25k iPhone app.  My PB was 35:47 over a year ago.  

I started following the5krunner.com  3 exercise "SMART" 5k training plan two months ago using the iSmoothRun app.  After the first month, I modified the B run to 20x 1:15/1 intervals at PB-40, and the C run to 6x 5/5 at PB-5.

When I run 1:15/1 intervals (at PB-40), I have great 5k times (a new PB, 35:44 yesterday even with the 1min walking recoveries).   When I ran straight for 6k a month ago my 5k time was really slow - 40:10.

After all runs, A/B/C or 5k straight my muscles do not feel tired but I do feel  sleepy the rest of the day.   This has been the case for the prior year as well.

Resting HR: 48
Max HR (Conconi test 10 years ago): 188  
Usual Running HR: level at 178 (95% ?)

"A: 2x 15/10"          Zone 2/3/4/5: 30%/12%/47%/10% (PB+30)
"B: 20x 1:15/1 run" Zone 2/3/4/5: 1%/11%/34%/53%   (PB-40)
"C: 6x 5/5"             Zone 2/3/4/5: 19%/29%/48%/1%       (PB-5)

(5k Zone 3 run/walk: 42:08)



On Thursday I will participate in my first public run, a 6k.  I am torn between running it in 1:15/1 intervals versus just running it straight with no ghost pacing for "it is what it is".

Any thoughts/advice anyone?