Self Driving Vehicles

I've wondered why Google is spending so much money and effort on designing a self-driving car.  I've also thought about what would happen in society if most cars became self driving.  I've come to the conclusion that the world will be a LOT better off with cars that drove themselves.  I'm sure there other big societal changes that would occur, but here are some that I've thought about, and all of them sound very good to me!

1. Passengers will be free to do more productive things while riding in a car, since they won't have to worry about driving, navigation, or traffic.  That could be reading, writing, interacting on internet, or taking a nap.  
2. No more cab drivers!  The people doing those jobs now, will be free to do much more productive tasks for society, whether in science, art, music, politics, or engineering.  Currently cab drivers have a difficult time making a living.  Since they often have to wait for an hour in order to get a fare, they then have to charge more to compensate them for their non-productive time in order to barely make a living.  A computerized cab does involve a slightly higher fixed cost, but requires no direct labor cost.  (Maintenance, cleaning, fueling, repairs etc would require labor, most likely)
3. No more DUIs!  Drunk or impaired drivers cause a lot of horrible accidents.  The costs to society of those accidents is often very high!  Injuries or loss of life to breadwinners or employees is a huge tax on society.  That is why police spend so much time involved with DUI enforcement.  
4. Traffic law enforcement could become a thing of the past.  Computers should be able to navigate roads without violating reasonable rules of the road.  Some rules of the road may even be changed to adapt to the new capabilities of computers.  For instance, vehicles on open highway could "draft" in platoons which will save energy.  Computers communicating with each other would allow vehicles to drive with very little following distance between them so cars behind the lead car in a platoon will save on fuel.  4-way stops could also be a thing of the past, if vehicles approaching such an intersection could adjust their speed so they would cross an intersection at alternative times without slowing down too much or stopping.  
5. Because DUIs and traffic law enforcement would be reduced, it would seem that the tasks for a police force would also be reduced. Taxes required to support the reduced police force could be reduced, or the current police force could be moved to more difficult law enforcement tasks. 
6. Driver training & licensing requirements may be reduced or eliminated.  Certainly some specialty vehicles used for construction or hauling may need drivers, who will need licenses.  
7. Auto insurance will be reduced or eliminated.  If computers are controlling vehicles, they should eventually be much more reliable than human drivers.  They don't get tired, and they don't get distracted.  If there are fewer accidents, insurance costs should go down.  If the concept for self-driving cars becomes a "fleet ownership" system rather than individual ownership, those corporations may need some sort of liability insurance.  But even so, there should be no real need for individuals to buy insurance against accidents. 
8. Vehicle maintenance could be reduced even further.  Yes, computers and sensors involved with the self-driving capability will initially have some flaws in hardware and software.  But as we see with modern electronics, they will eventually become very reliable.  If cars become all-electric, there would be fewer "moving parts" that could fail, and the overall reliability should increase.  That savings in maintenance cost, will trickle down to all residents in cheaper transportation.  
9. Improvements to Homes and Cities:  If self-driving cars became "common carriers" and not privately owned, there could be many further benefits.  Homes could be redesigned without as much garage space.  Cities could be redesigned to be more pedestrian friendly.  Cars could be parked in underground garage in a "last-in, first-out" packed configuration, which takes up less space than a design where everyone can get to their car in "random access."  

Skin Cancer Application for SmartPhone

The problem:

Some of the more recent information about spotting skin cancer has pointed out that one of the most important things to know when searching skin for cancer is how things have changed with time.  If spots or growths have changed, they deserve attention.  Many of us as we get older have many different spots on our skin -- all over our body.  It is very difficult for a dermatologist to spot a problem, even if all of the spots are looked at carefully.  For individuals like myself, who have seborrheic keratosis, searching all of our skin can take a long time.  There are too many objects or discolorations for any doctor to be able to remember, and too many for most patients to recognize changes, since changes occur slowly over time.

The Solution:   

Since so many people now have smart phones, I believe that an application could be built that would allow us (patients) to scan our own skin, map it, record it and then do automatic comparisons from time to time to look and identify changes.  The patient would perform a self scan of their body with the camera on their smartphone, where the resulting files would be encrypted and sent to a secure processor that would stitch the photos together, form comparisons and flag changes.  The resulting images and flags would be then made available to the patients dermatologist.  When the patient came in for an appointment, the doctor would be cued for possible areas of concern.

Advantages:  

• Patient does own scanning in privacy of own home
• Could cut down on time spent by dermatologist and improve accuracy
• Will help flag those subtle changes that could indicate skin cancer problems
• Forms a permanent record of a patient’s skin and shows changes with time
• When some objects are removed from patients by surgery or freezing, it allows doctors to clearly see if they come back
• This permanent record database could provide information for researchers trying to study types of skin cancers, discolorations, and growths

How it would work for patient:

Patient would download the application, based upon doctor's instruction, and then find a place with good lighting and follow vocal instructions provided by the application.  Instructions would be like this:

• While holding the phone in your right hand, Press the plus ( +)  button on your phone and slowly pass the camera over your left arm starting at the shoulder down to the tips of your fingers.  When done press the minus (-) sign button  Your scan should take about 5 seconds from start to finish.  
• Now do the same thing for the underside of your arm from your left armpit down to the tips of your fingers.  Press plus sign to start and minus sign to stop
• Finally, turn your left hand so palm is up, and scan under the arm past the elbow down to your wrist.

The whole process could take as long as a half hour to complete scanning the whole body.

What the application would do:

• Create a file for each of the scans along with metadata for the scan
• Verify that the lighting for the imagery is adequate for recording
• Make sure that the scan wasn’t too fast or too slow.
• Encrypt the file for privacy based upon keys provided by the server
• Uplink the file to a server

What the server and software will do:

• Capture all of the files uploaded and do a quality check to make sure the data is processable and is complete.
• Report in real time any errors, to allow the patient to repeat a scan if required
• Real time errors could include corrupted files, missing files etc
• Capture the patients metadata from the uploaded file and determine if there is a doctor to bill.  
• Stitch the files together to provide continuous skin coverage and align it based upon skin patterns (similar to stitching together landscape photos)
• Determine if the patient has a file from a previous scan.  If so, perform a comparison of the two files and identify changes based upon criteria developed from clinical tests:
• Tolerance for how much change is enough to be flagged, to prevent false alarms
• Tolerance could be based upon shape, size, and color of marking or object on skin
• If too many differences, software could self-adjust some tolerances and report the situation.  Could also determine if problem could be due to a photo registration/stitching problem
• Doctor may want to set a specific tolerance for change, depending upon the patient's age and skin type
• Extra filters may need to be enabled for patients who have other forms of skin problems, such as keratosis pilaris , hydrocarbon keratosis, or eczema.
• Inform the doctor’s office that a patient’s file has been processed and is available for review.

What the doctor’s interface will do:

• Doctor’s credentials and billing information will be validated
• Doctor will have a screen display showing the flagged locations where significant changes have been identified
• When doctor clicks on that spot on the body, the display will show a split-screen display with previous scan and current scan.  Doctor could opt for an overlay display and could click to see before and current scan.
• Doctor could also look back in archives if patient had multiple scans on file and perform comparison, or show time-lapse changes for particular areas of the patient’s body.

The world needs an Internet DVR

It is clear that television viewing as we know it is going to transition to an internet-based system.  Everyone wants video on demand.  We want to be able to watch the shows we want to see, when we want to see them.  We don't want to pay a cable company for channels we never watch (for example, non-sports people paying as much as $10/month on the cable bill for ESPN channels).
However, it is clear that the internet distribution system is now being presented with a huge "peak loading" problem.  I heard a statistic that 40% of all internet traffic is Netflix traffic at 8PM local times.  That probably doesn't include Hulu, You Tube etc.
At the same time, the cable companies are asking for extra fees from Netflix and similar providers to guarantee them "fast track" connection to their customers. The FCC is apparently going to approve that fast track service with some constraints.  Clearly, that fast track requirement is needed only for that "peak loading" time of day in each time zone.  The rest of the time, there is huge amounts of bandwidth unused.
The problem is that we don't have an easy way to "smooth" that demand.
We currently have TIVO devices that can record current shows off the cable and store them for future viewing.  Cable companies also have systems and cable-top boxes that do a similar store/replay service for viewers.  We don't have the same thing for digital media downloaded from the internet.  At least, I have been unable to find anything like it.  I think we need such a device! --AN INTERNET DVR!  Technology has advanced!  Mass digital storage is now very cheap, processor speeds are much faster.  We just need someone to build the device and sell it!

How would an internet DVR (IDVR) work?  The device will simply anticipate what shows a user might want to watch and download those shows during slower internet usage periods and store them on a local device in the viewer's home.  When the viewer decides to watch a show, it won't have to be streamed if it is already stored on the local device.  That will smooth the bandwidth loading on the internet, improve the quality of received video, and reduce the cost to everyone.

How does the device determine what to store?  DVRs for years have tried to learn a user's desires and anticipate what a user might want to watch.  Netflix has spent a fortune developing algorithms to help users pick their shows.  With a little more interaction with users, I think a smart controller could easily predict most of the shows a user might want to see.  In many cases it is pretty obvious:  If a viewer has watched 8 episodes of a series, it is likely that the viewer will watch the 9th episode!  I'm sure the algorithm for this device could continue to evolve, and would learn more about what each user is interested in seeing.

What are the problems?  

• Commercials: Providers of commercial digital shows on the internet want to guarantee that customers are forced to watch their commercials.  They have been "cheated" out of people watching them on regular TV and Cable TV because of the advent of the DVR.  Currently they need to carefully allow their data to be downloaded in such a way that commercials will be included.  It gets very cumbersome. --Possible Solution:  This new device needs to "honor" the content providers system, and protect the incorporated commercials.  Users, however, will want to be able to fast forward through them - -maybe there could be a reasonable compromise solution.  For example, allow users to skip duplicate commercials within the past hour, if they have "clicked" on the commercial when shown the first time.  
• Copyright Protection: Providers of "premium content" such as HBO, Netflix, and Hulu want to make sure that the shows aren't recorded and then redistributed for free to others.  They need to protect the copyrights of the producers of the media.  Possible Solution: All media is stored encrypted with a key held by the data provider for each show.  When the device begins to play the show, it exchanges its serial number identity for the key to decrypt that show.
• Indexing & Cataloging: Currently shows on the internet are like the "wild west" -- there is seemingly anarchy!  Some shows are available while they are being shown over the air, some aren't available for a week after it was shown.  Some are available on line for years, while others are only available for a month or so.  Possible Solution: The developer of such a device will need to negotiate an agreement with the content providers that the provider will give the device access to a  viewing catalog.  I would assume that the agreement to catalog the data would be easy to achieve, as long as the device manufacturer agrees to protect the commercials and copyrights as described above.

The more I think about this the more I think it isn't a matter of IF this device will be built, but WHEN.  I wouldn't be surprised if something like it is being developed or tested right now.  It could even be incorporated into the top of the line TV sets now.  

It appears to me that Apple, Microsoft, Google, Amazon, Netflix, Hulu, and TIVO all have vested interest in doing this design correctly!  The cable and telephone companies Comcast, Cox & AT&T probably would not want this to happen and would try to put stumbling block in the path of any progress in this direction.  The media companies and content providers would probably all gain from such a device, as long as their content were properly protected. However, they would resist it until they were confident of  the design.  The problem is that if each company develops their own system, we will have anarchy and none of them will be successful. There would be a lot of waste if we had another 8-track/cassette battle, or Betamax vs VCR fight.  If the developers got together and worked on a common standard, they could be accused of antitrust violation..  I think this might be something that will require some leadership from the FCC to establish some sort of standards for protecting commercials and copyrights, and facilitating indexing and cataloging.