Rafiki Update

I won't just sell the robot.

Imaging echo working with buddy and tie it into your email, schedule, siri (I use the same service for information that siri uses), add in watson and Google services that have just become available to devlopers for AI type abilities. Now add the most advanced facial recognition product available, and then put an ez-b on it for expandability. Add in the best voice recognition and speech recognition software available, and the ability to use ARC and the ezb or whatever else you want to add.

The robot prototype has cost about $3000 to build if you take out tools and other other expenses. You know how much mass production starts to decrease costs.

The power is in the software. I'll try to explain.

There are independent subsystems that are independently doing one task. These sub systems report back to and take commands from a primary controller. The ezb is also able to communicate to this controller. The software to run everything communicates with ARC. This could just as easially be any other robot platform because the primary controller is managing all of the systems that come with Rafiki. It would work without the ezb or ARC. There is just a hook into ARC.

The ai is a totally seperate system that is constantly running. The core of the ai runs on a server and is used to communicate to the rest of the world's data. It also contains the core data for all of the information specific to the installation. There are pods that are placed in different rooms. These pods all communicate with the server.

Imagine being in your bedroom and being able to ask questions or ask for your from door to be locked. Imagine using facial recognition or voice recognition to have the system change for you or for your child.

If you want to expand your system, add another pod. If you want to have your meeting transcribed, send it to the server and it will do it for you. Speak different languages and have it translated.

The robot is simply something that will attract attention. The power is in the software and in the ability to expand the system.

I am off to my weekend meeting with my investor and lead developer.

So far, Rafiki has its speaker, light ring, arm motors, neck motors, wheel motors, drop off sensors, Object detection sensors and lidar subsystems all functioning. The subsystems communicate with each other where needed. The cameras are working and facial detection and recognition are working great. The accel, gyro, compass and temp sensors are working great. Also, the ability to communicate with ARC is in place. The microphone is working from across the room even with background noise. EZ-AI is running on a server and the client is installed.

In 5 months, we have gone from thought to a 3 foot tall whole house robot platform that can understand questions and provide answers, along with understanding who is using the robot and change depending on who it is. It is very easy to add new sensors through the EZB and is customizable through ARC. This customization doesn't affect the core systems or how they function. This approach also let's you use very simple commands to use these subsystems.

I calculated it today and we have about 1500 man hours in the project so far to get to this point and have spent about $6,000 total on everything to build the prototypes. About $1500 of that can be considered lost due to waste or mistakes. This cost includes all parts, equipment and software used to make Rafiki a reality. There is still a considerable amount of programming to do, but this is the fun part. Making Rafiki come to life is the stage that we are at now. Good times and I look forward to the meetings this weekend.

My hope is to have some answers to some question provided this weekend and to be able to move forward with more funding to bring Rafiki to what I believe is possible with this platform. Being able to prove that what we have said is possible so far is possible will take us a long way to having funding approved to push things further.

Now to see how well Rafiki likes long car rides...

Good luck Dave

May The Force Be With You

On my way back home in the morning to work on some things and get some bugs ironed out. Anyone who has programmed anything knows that the first showing always has some unexpected things happen with it. The good news is that I have had some ideas on how to improve the design of not only the robots plastic parts, but also its sub controllers some.

I got to work with my son some and we were able to discuss how we would like updates to the EZ-AI server to happen. We also were able to discuss ways to disable features which are pay for type features to possibly provide a non-subscription option for those who are interested in that. It would require some more programming and would access it's data from wikipedia directly. This isn't our focus right now but it is being thought about. The spoken words would use some sst stuff that is available on linux.

We setup our server that will be used to authenticate subscriptions and services. This allows this part of the work to begin.

We discussed having different versions of EZ-AI that are more specifically suited to different types of industries and fields within those industries.

We started the EZ-AI plugin for ARC also this weekend.

Hmm, trying to think of anything else other than to say that work will be ongoing through the end of the year to finalize parts of the programming for the bot and for EZ-AI. We still hope to have videos up and start pre-sales of Rafiki and EZ-AI around the beginning of the year for product delivery hopefully by the end of 2016 for the Robot. I strongly believe that product delivery for EZ-AI will be much earlier that that.

Glad everything went well for you Dave. I'm really looking forward to EZ-AI. I believe this is one of the components very much needed today.

Just some changes that will help out quite a bit in working on Rafiki. Also, added a much stronger speaker and made some head modifications. Added some cutoff and power switch, along with 5 fuses to different things. Lets see, basically a rebuild from the ground up from what was posted before. The omni direction wheel in the back is much quieter than the caster I had in the previous videos... The head doesn't wobble when moving anymore. A lot of improvements that required a complete reprint and rebuild using lessons learned from the first build.

I posted a BOM in the public github also.

That music though......

Yea, he is from the same town as me. I like the words to this song. Not a country fan, but, the words have meaning for me.