Getting Started With the Propeller 2
BobCochran13
Posts: 10
Greetings All,
I'm very interested in learning what I can about the Propeller 2 and how to develop with it. Is there documentation I should read to get an overall grasp of the new platform? I think I should start playing with the Propeller 2 as soon as I can.
Thanks
Bob Cochran
I'm very interested in learning what I can about the Propeller 2 and how to develop with it. Is there documentation I should read to get an overall grasp of the new platform? I think I should start playing with the Propeller 2 as soon as I can.
Thanks
Bob Cochran
Comments
There is a draft document here
http://www.parallaxsemiconductor.com/Products/propeller2specs
and other documentation in progress as per the other threads in this forum.
The first engineering silicon is in the process of getting manufactured. In the mean time some of us are using Terasic DE0-Nano (Altera Cyclone IV FPGA) to emulate a propeller 2. That has several limitations - 60 MHz frequency, no analog pin functions, single cog (on DE0, or 6 cogs on DE2), however it does get us started.
Thank you very much. I looked over the specs and they are impressive, especially the video capabilities. I'm quite new to the Propeller 1 -- what got me involved in it is Jon McPhalen's SpinZone column in Nuts & Volts. I completed one of his projects successfully and will start a second soon, and that is my Propeller experience to date. I also toy with Atmega/Arduino and PICs but it is all "getting my feet wet" experimenting. I am an IT professional -- lots of application programming, but no embedded work. I think now is a good time for me to work more seriously with the Propeller.
Bob
It's a great time to get involved, with the Propeller 2 coming out. However there is an awful lot that can be accomplished with the Propeller 1, much of it is not documented in the manual as it has been figured out by users along the way. At the moment I am working with a 128 x 64 character colour text display that Kuroneko wrote.
If you are used to programming in C, have a play with SimpleIDE or Catalina.
What kind of applications are you interested in?
Yes, I agree the Propeller 1 offers quite a lot. What I want to do is to allow a friend of mine with very limited use of her hands to have the ability to operate a TV through voice command; and this should respond to her voice only (not to background noises or the voices of visitors or TV voices.) For example if she says "Channel up" or "Channel 10" or "Turn the channel up one" or "Down a channel" then software should process the command appropriately, perhaps through transmitting the correct infrared codes for what is wanted. I'd probably like to implement this using the CMU Sphinx libraries -- perhaps PocketSphinx.
I'd also like to be able to fly a small aircraft with a preloaded flight path that uses GPS for navigation. I'd like to rapidly load the destination and have the system figure out how to get to that destination and finally to navigate the craft there. For example: there is a duck house (a bird house large enough for a duck to nest in) on the lake which is about 1/4 mile from me. I would like the aircraft to navigate itself to this duck house and hover about 150 feet away from it and try to determine how many creatures are in the nest box. After doing this, the aircraft should return to my home.
Bob
Building a quadrotor to fly out to the duckhouse, take a few pictures and come back should be doable, but I doubt you'll be able to have it automatically count ducks. Also, you can probably get it to land somewhere on a landing pad, but not in an exact spot (At least not with just GPS).
Why do you feel the need to post in huge blue text? Very odd.
I have a feeling that if your aim is serious voice recognition (who is speaking) or image analysis (how many creatures) then you will be out of luck with the Prop.
These things are hard. Do you even know how to do them with the PC on your desk?
If you do, we are all ears.
Voice command will be tough on a Prop or Prop 2,(or ANY uC), but a Prop setup could do quite a lot for mobility limited users.
There are quite a few examples on here, of Prop PAL/NTSC displays, so you could apply a small LCD TV (aka car backing displays),
as a Remote Control GUI on top (or alongside) any fancier TV, as the Smart Remote Master.
Even Prop 1 can do this, Prop 2 will do it better...
Another Prop handles Pressure / Sensor to control transmission, at the operator end, and the rest is software.
ie you add Delay filters and Multiple triggers and larger display icons, depending on the exact disability areas.
Regarding the voice command, have a look at this Prop 1 project
http://forums.parallax.com/showthread.php/130825-Propeller-speech-recognition-robot
Regarding counting things, I would just aim to take a snapshot of the nest, and bring that image back to base for further processing. Hanno has created Viewport which has some image manipulation built in, and I believe can be extended. Doing the count on the PC probably makes more sense than on the propeller, although I'd love to see an attempt at it even if it takes a day or more to crunch a result. Roborealm may help and seems very good value.
good luck with it all
tubular
edit: big text. Can't hurt to have one thread that is easier to read!
I work with quite a few low vision people so I try to post in 14 point font, which is not at all huge. At my day job, I sit across from a blind person. My work group has two blind people. My second-level manager is low vision. I try to treat my low-vision colleagues with the greatest respect, especially after watching them struggle time and again with various devices to read text in small fonts. It does make me pause and realize what my emails to them might be putting them through as they try to read what I am saying. The blind folks use JAWS, which is not the most capable software, and requires heavy hardware investment.
I really don't know who reads my posts. The likelihood is that at least some low vision people use these forums. The #5 forum font is about 14 points. Blue is also an easy color to see. For this post I'll make a change and try black text. I try to post in a way that is inclusive of the possible audience, not just the "normal" audience. I don't think I should effectively turn away a low-vision person who has a great deal of interest in Parallax products by using the default font. I want to include everyone.
I know a lot of people in the different forums I participate in are somehow turned off by this -- there seems to be very deep, extreme dislike of my font choices all because I am not using the default font that the forum software uses. Some people go to great efforts to tell me how wrong my font choices are. Others tell me they are not going to respond to my posts if I don't use the default font size and color, period.
About your other question. Yes, I've done some experimenting with speech recognition on other platforms. I cross-compiled PocketSphinx for the Raspberry Pi Model B. I have done some work with it but not a lot. It doesn't recognize speech very well -- in fact the actual recognition rate is extremely low. Low enough to convince me that there must be some type of USB issue that causes issues with sound capture for USB microphones. Most recently I compiled PocketSphinx on an Intel-based desktop machine but have not had the chance to test its recognition with a USB microphone.
To me, the Propeller 1 is probably slower than Apple's A6 processor used on the iPhone 5. But not much slower, perhaps -- I haven't seen any comparisons and haven't looked for any. PocketSphinx also runs on the iPhone and so does some other speech recognition software, such as the Dragon products. It seems to me there might be enough horsepower to do at least some speech recognition work in the Propeller 1; the Propeller 2 certainly looks more capable in this regard.
So what I'm doing here is mostly just stating what my applications are -- speech recognition and flying about. I'm not saying I expect the Propeller to be a good choice for building such applications, but at the same time I don't have enough embedded experience to know anything in this area. I know there are some serious memory constraints for the Propeller but I lack the experience to assess the constraints. On the other hand, processing speech could be done with a realistic set of criteria scaled to both the goal and the hardware. Do I need to have an exact voiceprint match on the speaker's voice? Maybe, but probably not. If I simplify the criteria enough, it might be doable; I don't know. If the A6 chip can manage at least some speech recognition then perhaps there is hope that the Propeller can do the same. It seems possible. I've watched the Parallax videos on the Elev-8 quad copter and I think the Propeller 1 handles the complexity of radio controlled flight just fine. That's a great starting point. It can be pushed to do a little more, perhaps. The Propeller 2 could be pushed quite a lot more. I don't pretend to have any expertise. I do have a willingness to jump in and learn.
Bob
However, it's not necessary to actually post with a large font to achieve this. Every browser I know of can automatically increase the font size in incremental steps with a simple operation like e.g. ctrl-+ (control and the plus character). This has the added convenience of _all_ posts getting the same treatment. If different posts have different font sizes it just creates trouble instead.
Re: The font size. I do appreciate your efforts. My vision is not so good now a days either and I spend a lot of time with my browser zoomed in. On the various machines and browsers I have that extra large font size gives me problems. It's not that it's a bit bigger it's that it is huge compared to everything else. That makes it hard to read as there are so few words per line and the posts stretch down a long way. Further, the blue colour is horribly washed out on my old laptop, the resulting low contrast which makes for hard reading.
In general, the whole idea of HTML is that the final format is independent of the content. That is why we have various header tags, bold tag, paragraph tags etc. In the browser that content should be displayed however the user wants not how the author dictates. Of course a generation of web developers have subverted that idea by going out of their way to control formatting and get pixel for pixel the same results on all browsers and machines. That really does not help with the accessibility issues.
Re: voice recognition, etc. I don't think it is necessarily "non-constructive" to point out the obstacles and difficulties a person may face with an idea. Or that their expectations are not in alignment with reality. That may save them a lot of wasted effort and disappointment.
For example, when comparing a Propeller to whatever is in an iPhone it should be noted that by many regular metrics the Propeller not just a bit slower it is thousands of times slower. In terms of I/O throughput, memory capacity, memory speed and raw processing power there are orders of magnitude differences between an iPhone and Prop.
That's not to say that some kind of voice or image recognition cannot be done on a Prop/Prop II but one should be aware of the limitations and ready to accept the resulting restriction on what can be achieved.
http://www.trs-80.org/trs-80-voxbox/
I ran the UK users group and Tandy UK gave me one of the units which we loaned to members.
It used op amp bandpass filters and rectifiers with comparators. The outputs were sampled with a parallel input port and compared with values saved in a training session. It worked quite well.
Re: voice and image recognition - the easy way might be to off-load these functions to something that does these well. For example, have the prop run the quadcopter, and stream the video to a PC for processing. Or just strap an android device to the copter. The openCV guys have demonstrated how easy it is to do image recognition on a cheap not so smart phone. This is the path I intend to take, but of course I haven't finished my current project yet.
It does not HAVE to be hard, but it could be very elegant.