|
Project Overview |
|
|
Project Overview |
|
|
Project Overview:
The overall project is organized as 6 different individual projects. Each of these individual projects can stand alone as independent robotics projects and will be addressed individually. All of these projects will have an evolutionary phase where they might begin as something that appears to be totally unrelated to the finished Alysha project, but will in fact play a major prototyping role in the design of the overall robotic system. I will breifly describe below where I am currently at with each phase of this project and where I would like to ultimately take it. |
|
Alysha Body:
Current Status: Building the physical robot is one of the more challenging parts of this project for me personally. This is mainly due to lack of finances. I can't afford to just run out and buy custom made robotics parts to quickly toss a robot together. For this reason I'm starting out with some inexpensive robot bodies for the purpose of prototyping the navigation and motion of the robot body. The plan is that by ironing out all the details in an inexpensive body I will be better prepared to transfer this technology over to a larger more robust body when the time comes. In short, I can't afford to let the lack of a robust body hold the project up. So with that in mind I'm currently using very small robot cars pictured here for designing the navigation and motion control of the finished robot.  Final Goal: The ultimate goal of this project is to have a two-part robot base. A smaller indoor robot base that is driven by electric motors with rechargable batteries. As you can see it hasn't come along very far yet.  And a larger outdoor robot base that the robot can drive onto and use as an outdoor transport system. I'm considering having the outdoor base powered by an internal combustion engine running on bottled propane gas. And this outdoor base has even further to go before it starts looking like a robot base.  Top of Page |
|
Alysha Face:
Current Status: So far all I've done on this is find a useable face. I found this face as a party mask for $30. I took a chance on it and it actually turned out to be pretty nice. It's made of a material that is a lot like human skin, and is a full head mask. It's made to be worn to a party. So it's a full-size human head. It even include the ears on the sides. So it's a pretty nice mask to work with. I've also started working with Fusion 360 to start drawing up the plans for the internal skull structure. I plan on having the face animated using servos and stepper motors controlled by STM-32 boards. Final Goal: In addition to building the physical head software will also need to be developed to animate the head for speech. I did start a program in C# toward that goal. But I haven't worked on it for several years actually. Fortunately I still have the programs and some documentation that I created back when I was working on it. This animated speaking would need to be done in harmony with the Alysha Speak project, although this physical animation of the head really has nothing to do with what Alysha will be saying. This is just basically a robotic cosmetics to make it appear as if the head is actually doing the talking. Top of Page |
|
Alysha Arms:
Current Status: I actually have no idea how I'll be designing Alysha's arms. I would love to have arms that are somewhat human-like in apparance with an actual hand with four fingers and an opposing thumb. But to be quite honest that currently appears to be a goal that might be beyond realistic. Especially considering all the other projects involved in this robot design. What I have done recently toward this goal is to set up a tiny Me Arm and begin to program it to do various tasks. The Me Arm is extremely limited in what it can do as it has no wrist action in the claw as well as other severe limiations. None the less it has been helpful in learning some things about how to control a robot arm, including serving to allow me to write a program that can store arm motions on an SD card to be recalled later. This will come in handy for Alysha as she will be able to learn tasks and record them for later replay. In any case here's a picture of the current Me Arm project:  Final Goal: As stated above, the final goal is to have an fairly human-like arm that might possibly have a human-like hand eventually. What I'm currently doing toward that goal is to design a crude arm about the size and strength I'd like for Alysha's arm to be. I should probably mention here that Alysha will only be about 4 feet tall and to me she will always be a child. So she will have child-like limitations both physically and mentally. Top of Page |
|
Alysha Speak:
Current Status: I started writing a program a few years back using Microsoft Speech Platform. And even though I haven't worked with it in a while I've been refreshing my knowledge by going through my past notes. Everything is starting to come back to me which is a good sign. I recall being quite exicited about Mircosoft Speech Plaform because of their Grammar Builder. The Grammar Builder allows you to build a restricted grammar for the robot. Because of this the robot will only recognize words that are contained within the grammar dictionary that you have created. This allows for a great deal of flexibility as you can create different grammar sets for different situations. For example, when the robot is in the kitchen it can use the kitchen grammar dictionary. This would then allow the robot to focus just on words and terms that are most likely associated and used in the kitchen environment. Move to another location and the grammar dictionary can change. This also doesn't need to be dependent on location. You can simply ask the robot to consider a specific topic, and it can have a grammar set associated with that particular topic. So this can allow a robot to have a quite in-depth understanding of vocabulary without a need to have an overwhelming amount of words in a given grammar dictionary. Final Goal: The ultimate goal here is to have the robot be able to construct its own grammar dictionaries as it learns about new things. And this learning process should be very much like how a human child learns. In fact, I'm hoping to program Alysha to ask questions when she wants to learn new things. So hopefully she'll be like a little child and continually ask me why things are the way they are as she builds her grammer dictionaries. Note also that this part of the robot project can be done using C# on Windows without the need to actually have a physical robot at all. Top of Page |
|
Alysha Listen:
Current Status: There is a huge difference between speaking and listening. And this is also covered with the Microsoft Speech Platform Grammar Builder. The idea is to have Alysha identify with words she knows. These would be words that are already in her grammar dictionary. And those words will have associated definitions. So the idea is to have Alysha not only be able to hear words, but to actually know what they mean. This is possible to do using Microsoft Speech Platform programming. Final Goal: The final goal is of course to have Alysha be able to learn and associate various words and concept (i.e. the definitions of those words), into something meaningful to both Alysha and the person who is communicating with her. Top of Page |
|
Alysha A.I.:
Current Status: The A.I. part of Alysha's program will take place in the Raspberry Pi. This might seem like a limitation but it's really not. I have a lot of ideas for this that are loosely based on ideas of Marvin Minsky and his "Society of Mind" idea. And much of this will also be associated with the grammar sets that will be created using the Microsoft Speech Platform. I might add here that this larger architecture is not part of Microsoft's Speech Platform. I'm simply using their Grammar Builder as a tool. However, the way that I am employeeing it really has nothing to do with Microsoft's Speech Platform itself. Their platform is simply a tool that I'm using for my purposes of A.I. Final Goal: The ultimate goal is to get the robot to be able to make meaningful connections between concepts. Does this mean that the robot will be "conscious"? Well, that can all depend on what you definition of consciousness is. For me, all it will mean is that the robot will be able to make connections between various concepts. That doesn't mean that it is having any sort of experience. It's just making connections between concepts. And that's the ultimate goal. Top of Page |