House Basketball - The pitch

Turn the odds against the house with House Basketball.

House basketball - The fun part

The computer vision part is pretty interesting because it involves reconstructing the players’ positions and could move into understanding how posture conveys feeling. The analysis of the computer vision data would also be quite interesting.

House Basketball - The growth potential

I don’t think this has much growth potential in terms of number of clients. The best way to get growth is to move into doing this computer vision and analysis for other sports. It could grow to be a multi-million dollars per year business.

House basketball - The monetization

This is a product whose value decreases with the number of people who use it, so it really may be limited to a few dozen people. It might be best to set this up as a revenue share plan, which makes more sense than charging $100,000 for a piece of software.

House basketball - The idea

Most gamblers often only have data on the wins and losses and score differentials of the teams that they bet on.  There’s much more data available that could improve betting strategies and outcomes including point-by-point summaries of the game. That is what professional gamblers currently have access to, but there’s even more out there if you know how to use computer vision. With a few different camera feeds, it’s possible to reconstruct every player's exact position on the field at every second. These can help answer how many contested shots a star player has been making and how much his teammates have been giving good picks or dribble penetration before kicking it back to the star. With more advanced computer vision, you could see how players’ posture changes when they are up or down and whether they slow down when they are behind by a lot.

With the data the software makes available, it could be possible to accurately determine tendencies of the coaches, such as when they make tactical mistakes and whether they know how to recover when an opponent spots a weakness in their defence. The data could also figure out when refs make mistakes. This could be useful if a ref tends to call a lot of charges and one of the star players for the upcoming game pushes the envelope with charging. There are many small but important things this software would help uncover.

House Basketball - The motivation

Sports betting is big money. Being 10% better at predicting the outcome of sporting events can turn a hobby into a career. Software that can watch games for you and give summaries is clearly a good investment for professional gamblers.

Medical Summary Software - The pitch

Physicians should not have to spend hours a day looking up patient information, computer software should give summaries of everything they want to know so that the doctors can focus on delivering great care. Medical Summary Software does exactly that.

Medical Summary Software - The fun part

The key to this software it something like natural language processing. I say “something like” because it involves parsing medical documents and texts. The same principles from NLP apply, but the content is quite different than books or blog posts. Pushing NLP to tackle the language and forms of a specialized field could be quite challenging.

Medical Summary Software - The growth potential

There are almost one million practicing doctors in the US. If 10% of them signed up to use this software, it would generate around $20 million a year in revenue in the US. If this went global, it could grow to many times that.

Medical Summary Software - The monetization

I think the software would be sold on 3 year licenses for around $600. I think the software would create ten times that amount in savings.

Medical Summary Software - The idea

There is (finally) a push to get medical documents digitized. Digital medical documents facilitate a number of interesting ideas including doing data analysis to connect genes with diseases and using medical history to predict future problems. In addition to those large (insurance) uses, there are smaller efficiency-oriented uses for these data.

Doctors and other medical professionals need to review patient files before seeing them. Looking over a patient’s medical history can take a lot time or, more realistically, take a moderate amount of time but a involve a lot more skimming. What these doctors really need is a summary of the relevant parts of the patient’s medical history given what the patient has come in for. If the patient is complaining about chest pain, any previous heart problems and cholesterol levels and cholesterol drugs would be clearly relevant, hearing loss probably would not.

The software should be an aid to doctors to also look for hints of other things in the patient’s past that might be related. For example, the doctor should be able to get a summary of everything in the patients history related to shoulder injuries in case the chest/shoulder pain is from an old injury and not pain from a heart problem. The process should be fast and interactive where the doctor can enter and refine queries over seconds. Faster interactive summary software could make doctors more efficient and more effective at diagnosis.

Medical Summary Software - The motivation

Doctor’s have to review your file every time you come in. It would be nice if a summary of your entire (relevant) medical history was available on demand.

The Elf Shop - The pitch

Is gift giving just a little too stressful right now? Take a break and let The Elf Shop do the hard stuff. All you have to do is choose how you want your presents to look and we take care of the rest. We’ll even send you a picture of how your presents look before we mail them.

The Elf Shop - The fun part

The real puzzle here is how to organize the workers and the storage of incoming packages to be efficient. Also, there is one time of year, Christmas, where the demand could increase by a factor of ten or more. The operations would likely need to be temporarily scaled-up  for the holidays.

The Elf Shop - The growth potential

So, I couldn’t find any numbers on Christmas revenue or birthday presents, so this is going to be a little rough. There are about 300 million Americans. Only about 10% of the population would really want to use this service. So if about 10% of those people used The Elf Shop once a year for their gift giving and The Elf Shop made about $10 profit per shipment, that would be 30 million dollars a year.

The Elf Shop - The monetization

The customer pays for the wrapping costs for each of the presents as well as for the shipping to the final destination. There would be a wide selection of wrapping paper, bows, ribbons, bags, tissue paper, etc.  There could be much more revenue if the aim is to be high-end. The nicer the wrappings, the larger the margin could be.

The Elf Shop - The idea

You buy presents from online stores, and even some brick-and-mortar stores. All packages are sent to The Elf Shop, where they will wrap all of the presents and then mail them directly to the lucky recipient. This saves the shopper time and could save some money on wrapping supplies.

The Elf Shop - The motivation

Amazon and other major online retailers are typical starting points for shopping for gifts. Amazon has (passable) gift wrapping. But what if you buy something from two different places? Your options are to mail the packages directly and hope they get there on similar days or have them delivered to your house and then you have to wrap all of the presents then mail them off again, which can be a big hassle. That’s where The Elf Shop comes in.

Virtual Robotics - The pitch

Both students and professionals struggle with getting robotic systems to work. The key to quickly iterating on an idea and getting it off the ground is to simulate it first. This is exactly what Virtual Robotics is designed for.

Virtual Robotics - The fun part

Well this is easy: it’s Robotics! Robots are cool, and robots that don’t have glitchy behavior because I can’t solder very well is even better.

Virtual Robotics - The growth potential

I think there are thousands of college students a year  (and even some high schoolers) that take robotics courses. There are maybe ten times as many professionals who would use this software. Between the two groups I think this product could make millions of dollars a year in revenue.

Virtual Robotics - The monetization

The software would be a cloud service and come with a monthly license fee of $10 per month for the professional version and $5 per month for the educational version.

Virtual Robotics - The idea

I think that there are two use cases for good robotic simulation: education and prototyping.

Education

Robotics is fun, but it can be challenging to get into. It’s not nearly as easy as picking up a python book and making cool starter programs. It would be better to separate the hardware design and the software. With a virtual robot, a student could learn how to read data from the sensors and drive the servo motors. They could also see what happens to their virtual robots running their code in different environments. Having detailed debugging tools could allow the student to see exactly why their robot got stuck in the corner and kept running into the wall. After the student got a handle on the basics, they could put their virtual robot into more realistic environments where they could experience wheel slippage and sensor data degradation.  Running in more realistic environments could allow them to slowly learn to write robust software and even make unit tests for the robot in specific challenging areas. The last piece is to allow the student to modify the hardware in their virtual robot and see how their code would work in a slightly different robot or what more fine-grain sensors and motor controllers could allow their robot to do.

Prototyping

Serious robotics companies and teams have custom software to simulate their robot behavior. Many of them also have hardware-in-the-loop simulation. The software that I’m proposing here isn’t fundamentally different than what many already have. The biggest difference is that Virtual Robotics would be a flexible simulation framework where companies could tailor it to their needs be them aerospace, ground based automatic vehicles or even those that swim underwater. This would make developing custom simulation software from scratch unnecessary.

Virtual Robotics - The motivation

Robotics is really hard. You have to do hardware design and software design and each of these can impact the other. It would be nice if you could do both the hardware and software design in the computer before building real prototypes.

Digital Cane - The pitch

Having a vision impairment isn’t safe and it can make life really hard. Digital Cane helps by making everything around the blind person as simple and easy to process as a sighted person. Also, its head-mount camera makes picking up the subtleties of conversion a breeze.

Digital Cane - The fun part

Like my idea for helping the deaf, the fun challenge here is to get the right pairing of hardware and software that ‘just works.’ This should be like a new set of skin for the blind person. Ideally users would use the word ‘seamless’ to describe the product.

Digital Cane - The growth potential

There are over 1 million legally blind people in the US. If 1% of them buy one of these units per year that would $30 million in revenue per year.

Digital Cane - The monetization

The software and hardware would be sold as a set. I think it could sell for $3,000 per unit.

Digital Cane - The idea

The motivation

Life as a blind person is fairly challenging. Tools like guide dogs are nice, but it’s still hard to have the same quality of life as a non-impaired person.

The idea

There are two big areas where being blind is a big impediment: navigation/hazard avoidance and personal communication.

Navigation

Navigating streets and buildings can be dangerous for the blind; cars and stairways can be extremely hazardous. What blind people need is technology that allows them to use another sense to replace sight. I think the Microsoft Kinect sensor is nearly perfect for this job. A few well-placed sensors could measure the distances to most things around the blind person. Ideally, the size of the sensors would be smaller, because wearing four Kinect sensors could look rather silly. Nevertheless, these sensors would collect everything the blind person needs to traverse safely. The data from the sensors needs to get conveyed to the blind person. I think the best way to do that is a band around the chest or stomach. The band would have thousands of cell phone vibration motors. The motors would vibrate when there is an object close in that direction. The closer the object the more the vibration. The sensors would be angled at a slightly downward direction (towards the ground). This would result in the vision impaired person normall feeling a vibration when even ground is ahead and different vibrations when there is a staircase or other obstricle ahead. The Kinect sensors may need to get paired with longer range but coarser sensors that can ‘see’ cars. A special warning, perhaps auditory, should be issued when a fast moving object is approaching and might hit the user. I think with these sensors and the vibrator motors, blind people could see well enough not to need a stick or a guide dog.

Personal communication

Blind people are at a disadvantage when it comes to in-person communication. People can say as much with their posture, facial expressions, and gestures as with their words. To reach parity with people who can see, the blind need help. The Kinect sensor above could be useful to detect gestures, and maybe posture, but it couldn’t help with understanding people’s facial expressions. I think something like Google glass or some other small head-mounted camera would also be needed. Facial recognition and emotion recognition are now possible.  So, the user could know who the speaker is and what their face says about their mood.

The monetization

The software and hardware would be sold as a set. I think it could sell for $3,000 per unit.

The growth potential

There are over 1 million legally blind people in the US. If 1% of them buy one of these units per year that would $30 million in revenue per year.

The fun part

Like my idea for helping the deaf, the fun challenge here is to get the right pairing of hardware and software that ‘just works.’ This should be like a new set of skin for the blind person. Ideally users would use the word ‘seamless’ to describe the product.

The pitch

Having a vision impairment isn’t safe and doesn’t make life easy. Digital Cane helps by making everything around the blind person as simple and easy to process as a sighted person. Also, its head-mount camera makes picking up the subtleties of conversion a breeze.