Linux based systems are nice for power users, but have also major issues, like there is a much larger learning curve to get stuff up and running, versus a simple arduino environment. Linux systems can be slow to start up, some as much as a minute… However once you get over these hurdles, there is lots of power to be had and lots of great stuff available. Some of the boards I am playing around with include the Raspberry Pi, Beagle Bone Black, Odroid U2 (and today U3) and an Intel NUC (too expensive for hobbyist).
The new Galileo, Good question, Will be interesting to see how well these perform. Personally I am more interested in the Arduino Tre. Not sure when they will come out, but they will be like having a BeagleBone Black with a Arduino Leonardo connected up to it. Sort of like the Arduino Yun but more.
Arduino Due, These are interesting. I have one, but have not done much with it in quite awhile. Yes, they have more power than the Megas, but they are only 3.3v devices. Also like most megas, they are rather large and bulky and don’t fit well on smaller robots.
While I am mentioning it, I have also played around with several other boards as well, including ones based on Pic32’s like the Chipkit Uno32 and Max32. Again these are great boards, but again 3.3v. They have their own version of the Arduino IDE. Not sure if they have been able to integrate yet into the Arduino 1.5x platform or not. I was waiting until they were at least Arduino 1.0 compatible before I played with them again.
This week, I am playing with two different platforms. The first is a Teensy 3.1(forums.trossenrobotics.com/showt … Teensy-3.1
If I like this board well enough, I may do a quick and dirty board design, that adds power stuff, IO connectors and an XBee connector, to make it easier to add to different robots…
This is the new Odroid U3 board. Note: I put an Arduino Leonardo in the Photo to see the size. Also in the photo is their new IO shield for the board. It uses I2C plus a UART to talk to the main board. The I2C has an IO extender on it which gives you 16 IO pins, the Uart is connected to an Atmega 328, that has an Arduino bootloader on it, so you can directly program it from the U3 board and it has the 20 IO Pins. I am not sure yet which robot I will put this one one. May go on a Rover for me to learn and try out ROS… More later in other thread.
Of course there is going to be a longer learning curve for GNU/Linux than for Arduino, but you are comparing oranges and pickles here. GNU/Linux is a full blown operating system, and the Arduino IDE is just another application that can run under Linux, Windows, and MacOS. However, as one gets deeper into doing stuff with an Arduino, there is more learning curve to go through, such as when one wants to start writing their own libraries.
Start up time can be an issue for any operating system, depending on how it is setup and how much stuff you have starting up at boot time. This is not a hurdle to be gotten over. It’s entirely dependent on how an OS is configured to start up.
This could prove to be an interesting combination. I’m still on the wire regarding the Yun, but have to admit I am very curious about it. I might have more interest in the Tre, so will have to look at it more closely. Personally, I think the UDOO folks have the right idea, and it really looks like they have gotten in right. I’m very anxious to get my hands on an UDOO quad.
Any board can be designed to be 5V tolerant, or even dual voltage tolerant (at increased cost, of course). With more capacity and capability, you are going to have to deal with a physical increase in board size at some point. There just isn’t any way to get around this, because you can only fit so much into a given amount of space. You have to pick the right board for a given robot, or, you may have to redesign your robot to handle a larger board.
Again, the power issue is not an issue at all. It’s not like it was when manufacturers first started making 3.3V stuff. There is more and more stuff working at 3.3V all the time, and now we are seeing a move to 1.8V stuff (as it was when 3.3V first started). It’s just a natural evolution of electronics to lower power usage. I remember not liking 3.3V at first, but now I don’t have issues with it. Lower power consumption is great for robotics!
This would be an interesting thing. Small board, large capability. I like that! I’ve also looked at these. Unfortunately, I can’t afford to get all the boards I would love to tinker with. I have a long list of boards I would like to have to tinker with. There are also other interesting boards on the horizon, such as the tiny BLEduino (Bluetooth 4.0 enabled Arduino), and also tiny Spark (WiFi enabled, ARM 32-bit Cortex M3).
This is a big goal for me. I want to get a platform that I can easily install and run ROS on. I’d love to experiment with such a thing on W.A.L.T.E.R. Right now, I think my minimum capability set for running ROS would be an UDOO quad, or better. I really like this apparent new trend of blending a Linux system with Arduino compatibility. You are so fortunate to have several robots to tinker with. I really think my next board will be an UDOO quad, which will go right on to W.A.L.T.E.R. for sure.
Hope you don’t mind that I migrated the last couple of postings to a new thread. Probably should not pollute the announcement thread.
Dale,
You actually made my point about Linux. I think it is great for many of us, but I personally don’t think it is a starter system for someone just getting into robotics. Unless of course they are nuts like myself However how many of us know enough to reconfigure the Beagle Bone and/or RPI to remove everything not absolutely required to get it up and booted? Note: there are those up on the BeagleBone forums, who think all of us who use them, should fully understand how to build and update the kernel and if you run into problems, you should be able to diagnose and build your own patches to make it work…
But like you said, in order to get Linux to do what you what you want it to do, you have to know how to configure it… There is something to be said about having special purpose hardware that just starts up and runs your stuff and for the most part nothing else… Like the Teensy 3.1 is an ARM based system, that boots quick and is ready to go. Other manufacturers are thinking along this line as well. Example Robotis has come out with a new controller: robotis-shop-en.com/shop/ste … umber=1008 I will probably try one of these out soon…
I also agree with you about the 3.3v versus the 5v, that the industry is migrating that way, which is great. However I do have concerns on the Arduino front again for the near term. Example I have an arduino and I have shield X. I now get an Arduino Due and plug it in and fry the Due as the shield was not made for a 3.3v setup. Yes there is a newer shield definition to address this, but I don’t know how many shield manufactures connect up to the new Voltage Reference Pin and properly use them… Again long term great… But personally I am happier when a board manufacturer takes this into account, like when Teensy went from their 3 to the 3.1. The new 3.1 is protected. As for size, the board is darn small
This brings us back to the learning curve of Linux (or anything new). Even I cringe when I start considering building a custom kernel, and I’ve done this sort of thing many times on various platforms. Yes, Linux requires one to have more knowledge of the system to use it to best advantage. It’s definitely a longer learning curve. You have to get familiar with the package manager for your distribution, or you won’t get far with Linux. I don’t go deeper into Linux than I absolutely have to, and I’ve been using it since before kernel 1.x.
I had a real fight getting I2C working on the BeagleBone Black. This isn’t because I’m not capable when it comes to Linux. It’s because the information I needed was spread all over several websites and some of it conficted with other information. It was difficult to sort out what was correct and what I actually needed to do. It turned out to be much easier than I thought, after I finally figured out what I needed and had followed the correct instructions.
I am not beyond using a special purpose board if it gets me to where I want to be quicker and easier. I can do the “black box” approach. I need something that has two I2C buses, is small, fast, and can handle whatever processing for sensors I need to do. Being Arduino compatible would be a big plus. I can’t use SPI on my BotBoarduino because I need the pins for other functions. Oh yes, and the SPI and I2C pins can’t conflict with analog inputs or outputs, or other specialized functions like PWM. I am going to have to take a closer look at the Teensy 3.1.
I’ve been there, done that, got several t-shirts. I still don’t have much sympathy for somebody who does this and burns up their nice shiny new board. It’s called RTFM all the way. You need to read the documentation available for your shiny new board!
Again, you have to check the documentation and instructions for new hardware. It’s the manufacturer’s responsibility to make all required documentation and information about their products available so users can use them correnctly. It’s NOT their responsibility to hold your hand while you are learning about how to use their products. They do need to have some way for us to communicate questions to them if we are uncertain about something.
I think the Arduino Due is the future for the Arduino line. The Arduino people have gone as far as they can with the Atmel chip line, and they need to upgrade. Arduino can go a long ways with the ARM line, and they don’t have to touch Linux if they don’t want to. There are many great ARM7 chips - one of my favorites is the LPC2000 line from NXP.
Their stated goals is that is it an open-source prototyping system that is flexible easy-to-use hardware/software for artists, designers and hobbyist…
I do believe that boards like the Leonardo and Uno will continue to be a very popular boards, for many smaller projects.
I think the Arduino Due is pushing this for them, and they may be having growing pains from it. They are still in beta on their IDE for this and my impression, is one of the next betas for it may completely rework (again) how the libraries work. How do you define libraries that work on the Due but not on the Atmega based Arduino boards? Can you have some that work on both? …
As for outgrown the Atmel chip line. Nope: They for example could have gone to to 32 bit Atmega chips, likewise they could have gone to XMega chips. For example I am pretty sure the boards like Roboclaw use the XMega chips…
Also not really sure from their more recent stuff what they are thinking, for long term, as for the higher ends, they have the Due, followed by the Yun (Atmega 32u4 plus OpenWRT), Tre (Atmega 32u4 + BBBK), Galileo(Intel chipset)… Let alone others based on Pic32 (Max32) Will be interesting to see what happens next
I am also torn, that once you have the horsepower, how do you use it? Do you still keep it like a real simple Arduino app, that has only the main processing loops, plus interrupt handlers? Or do you start to add more advanced things like Threads? If you add threads, then you probably need other more advance things like Mutex, or semephore… How much more can you add before you lose the easy to learn/use idea? If you go far enough, then you might as well go to some form of Linux… And when you cross this type of line, do you stay with an 80 or 96mhz chip or go up to something like the Raspberry Pi/Beagle Bone or Odroid, which the prices are in the same ballpark, but run at much higher speeds (700mhz+) or example Odroid U3 (1.7ghz quad…)
Then there is the issue of adding IO capabilities. Again multiple valid ways to go. Could go like the BBBk or RPI and have quite a few IO pins available to you on the board, which are great. Or like the new U3 board, there is an expansion board that adds quite a few IO pins. But another valid way with many of these Linux boxes is to use USB for it. Example: There is a simple (very small) board you can plug into the USB port, to run AX-12 servos. Likewise you can plug in the new Roboclaw into a USB port… For sensors you can plug in other boards to handle this, including a simple Arduino board. Again many ways to do this…
As for what Lynxmotion should do? It really depends on where they are wanting to move the product line toward. Not sure if they are going to evolve to other servo lines? Will they be actively involved in developing software or mainly relying on community support? … A linux based system would be great. Especially if they come up with a standard release, that comes pre-installed on the system, that supports their product line… But that is a lot of work and would take a lot of support. Again not sure where they wish to go.
Yes, of course the Uno and Leonardo boards will continue on. They are very appropriate for many projects. I will still use my BotBoarduino and Decimila, even if they end up not being appropriate for use on W.A.L.T.E.R. I always prototype first with my Arduinos, before I move things to more advanced platforms. The first Linux based platform I use is my Raspberry Pi boards - I believe it is the Arduino of the embedded Linux systems.
I think the Arduino folks are going through growing pains, in general. This can be seen by the wide variety of boards they are bringing out now. You can build multi-platform libraries, but it does add complexity for the developer. I also believe this added developer complexity can be hidden from the end user. It’s not the end user’s responsibility to adapt to what the developer provides - it’s the developer’s responsibility to make things as easy as possible for the end user to use.
Yes, they could have done that. However, they didn’t, and there must be a good reason why they didn’t. This is a question I would like to hear the answer to. I will have to take a closer look at my RoboClaw 2x5 the next time I have W.A.L.T.E.R. upside down. Personally, I don’t really care what chip a given product uses as long as the end product does what I need it to do and works the way I need it to work. I am extremely picky about what boards I buy, and it took much thought before I settled on switching to the RoboClaw line of motor controllers.
At this point, I think it’s impossible to tell what the Arduino team is thinking. I also don’t think you can put an order to where their current products fit. Each new product is too different, and fits different kinds of uses. The Due keeps in line with the traditional Arduino layout, which I think is excellent and one big reason I want to get an UDOO quad.
The Yun is intriguing, to say the least - it almost seems like a little brother to the UDOO. It’s interesting enough to me that I want to get one fairly soon. This might be the board I want to put on the little SES balancer I am working on.
This depends entirely on what kind of application is being worked on. It depends on what capabilities a developer needs to make their application work to accomplish its end goal. I only add the level of complexity I actually require to do what I am trying to do. I’ve known for a long time that W.A.L.T.E.R. would need more than one processor, and that one of those would be Linux based.
As a board becomes more complex, the developer has to work harder to mask that complexity and still keep it easy to access and use. I frequently add a layer on top of a library I am using in a project. I want somebody looking at my code to understand my application, without having to consider what the underlying libraries are doing. This often is a lot of work, but has always brought back a lot of gain in the end.
This still comes back to choosing board(s) with the capability you need for a given project. Then you must deal with whatever level of complexity the board(s) have, and wrap your code around that. It’s the designer’s job to choose the right components for a project. When we decide to build a robot, we are the designer, mechanical engineer, electronics engineer, software engineer, and programmer/coder. We do the equivalent of what entire teams of people do in organizations and companies - just on a smaller scale. My vision goes far beyond just building W.A.L.T.E.R. 2.0, and W.A.L.T.E.R. 3.0.
Yes, of course! I believe the best boards for robot “brains” are going to be those that combine some form of Linux/UNIX with an Arduino Due (or other Arduino product). Linux is not capable of handling things in real time for the most part. Yes, I know there are so called hardened and real time Linux kernels, but they are still limited in how fast they can react to events. This is why I believe we need Linux + Arduino (or similar). We can code time sensitive stuff on the Arduino type board, and the Linux part can communicate with it, or the Arduino type board can send an interrupt when something needs to be handled. Whey I use the term “Arduino type board,” I mean a board that is baremetal programmed and does not run an operating system.
I have not tinkered with using a USB based board physically connected to a Linux based board via USB. The way Linux handles USB ports bothers me, because your device can change ports without notice. This might be solvable by writing UDEV rules, and I need to learn more about how to do this. My original idea was to do exactly this - have a Linux based board with one or more USB connected Arduino type boards to handle specific tasks. There is also the possibility of using a bus such as SPI or I2C to connect boards, but in my case I don’t have any pins available on my BotBoarduino (all but two pins are allocated) to add SPI, and the only hardware I2C port has sensors connected to it, which would require the BotBoarduino to be a slave. I have not yet investigated multi-master I2C.
What direction any company, including Lynxmotion, should go depends on what their customers and users want and need, or at least this is how it should be. I expect Lynxmotion to be responsive, like they have always been. The BotBoarduino is an excellent example of their responsiveness, and it’s one of my goto boards, for initial project design and development.
One thing I think Lynxmotion should seriously consider doing, is creating shields for existing boards such as the Arduino Due, ChipKit Uc32 and Max2, and others. They really haven’t explored this much, even though I think they were (or maybe still are) considering making a shield for the Seeed Studio Arduino Mega work alike board. Lynxmotion is extremely good at taking technology, making it easy and convenient to use, and bringing it to everyone. They should capitalize on this ability!
My W.A.L.T.E.R. 2.0 is a Lynxmotion based robot through and through. I designed his decks specifically to be compatible with SES brackets. Even his color is an homage to Lynxmotion, because many of their deck components were made of yellow lexan plastic when I first started out here. I have the greatest respect and admiration for Jim Frye and his family, and RobotShop has VERY LARGE shoes to fill.
Good question, but I don’t know if this can be answered as at least for myself this depends on a lot of things including what is Lynxmotion’s vision for the future.
Are you going to stay with Hitec servos? Or are you going to go to other ones like the Dynamixel or Herculex or …
Are you aiming for the entry level hobbyist or more advanced hobbyist or education or ???
Are you talking about working and fitting with current robots or going to design new robots that fit… Example: Boards with the standard Arduino Mega form factor did not fit with many of the current robots. Yes on many you can tack it on, but that is not always what is desired. At one point worked on shield for board geared more toward Seeeduino Mega form factor, but decided not enough extra bang for the buck…
SBC - Lots of pluses, not sure in a year or two if Linux or Droid or Firefox OS, or some of the others will come out working better. For example if you look up at the ODroid forums, they have both official images (Both Linux and Android) done by the company itself, but in addition to this, they have other images available, that were done by some other people, where the images are specific for different things: Android (for entertainment, full blown…) Ubuntu (different versions 13.10, 13.05, 12.11… - Ones set up for code monkey, ones set up for ROS with OpenNI and OpenCV)…
Personally some simple Arduino(ish) board is probably the safest bet for the near term for new to mid users. (Don’t see a full linux machine on a Brat).
But for higher end setups I think it would be great if Lynxmotion had a preferred SBC with an image setup, that had all of the stuff that was needed. But to get there I think you need to have people in house driving this (hopefully with community support and input). Example Trossen has done a lot of work toward Mechwar with Andrew, I see DR Robot has a ROS setup… There is Turtlebot setups out there… (Also sort of like having Eric there to drive UAVs)
As for Tablets and Smart phones. I personally think these will great input devices for different robots, not sure if I would use them for the main brain. But that is me.
Sorry for asking more questions than giving an answer.
Feedback and questions are always appreciated, and we’ll disclose what we can.
What we are seeing is that single board computers which can run a complete operating system (usually Linux) are dropping in price to the point that customers who are new to the field don’t necessarily know how to proceed. The Raspberry Pi is a good example in that for $25 you get a complete SBC capable of running several versions of Linux, complete with video output and more, whereas a normal Arduino costs $25 as well.
RC servos were what allowed the Lynxmotion kits to be designed and built in the first place, and they will always likely have their role in robotics. This having been said, there is technology available, at the right price, which adds considerably more functionality than R/C servos. We plan on jumping to the head of the pack again soon More details to come.
Without going into specifics, we can say that we intend to maintain the current customer demographic
Should not matter too much; if a board from a third party is to be used, it’s easy to design and build an adapter plate so it’s compatible with current models.
If this were to be the case, what type of functionality might people want? For example:
Hardware capable of handling image recognition and object tracking?
Ablility to use it as a stand-alone computer?
BT / WiFi integrated? HDMI port?
Sensors built in?
Basic board and shield concept?
Able to handle 3D sensors?
etc.
The idea of a specific “Lynxmotion image” would certainly be appealing as it saves considerable setup time.
Are there any SBCs or microcontrollers you have simply fallen in love with, or offer far more value than the others, or could potentially be the best with some modifications? Alternatively, there is always the approach to create something from scratch.
Open to any and all feedback, comments and questions.
My guess is for some of the stuff mentioned here it would be better to talk about offline… As I have said, I am torn here. I personally don’t think most of these Linux SBC boards are good for entry level setups… But I fully understand the cost benefit issue. These boards give you a lot more horsepower. As for building your own board, If you were going down that path, I would probably try to clone one of the existing open source ones, with which ever features you wanted… But from I have seen from the different boards, this would take a whole lot of support.
Also cost to setup a system, can be a bit confusing.
With most of these boards, you need an SD card to work. Needs to be a high speed one as well… Some have emmc which is faster but more expensive.
With these types of boards, you need network capability. Being that it is on your robot, you probably need/want for it to be wireless. So you need to add the cost of a network adapter.
With many of these boards, you will find you want more USB connections that there are ports, so you probably need an external USB hub (at least I do). Some of these probably need to be powered hubs. Example Raspberry Pi has problems if you draw very much through it’s USB port. This happens with some wifi adapters for it.
Most of these boards can only handle 5v (< 5.5v) power. As this is a robot, that you wish to run without wires and your motors/servos… Probably run on batteries that have higher voltages (6, 7.2, 12v), you probably need some form of voltage converter.
Most of these boards require you to set them up plugged into an HDMI display with a mouse and keyboard (at least to start). So you need those to begin with. Some of these newer boards have micro HDMI port, so if you are like me you need to purchase one of these as well.
I personally like sound on my robots, So I purchased some form of small speaker. On some boards, can plug directly in, on others the only sound output is over HDMI, so purchase a USB sound board
So again cost is an interesting question.
As for which of the boards I like and why:
I have 2 RPIs - nice cheap boards… But the slowest of these, also have had issues of SD cards corrupting.
I have Beagle Bone Black - I really like this board, Lots of IO, like the ideas of Capes (Shields). But earlier on was real frustrating as trying to get simple things to work, did not turn out to be simple. Like configure a set of IO pins to enable a UART. Also updates caused the system to die. It is getting better now, as they are moving to Debian release and simplifying the IO stuff. Still had issues with cheap Wifi adapter dropping connections… Changed to a different one, works better now. Has 2gb emmc on it, which is nice, but still need micro sd card to update.
Odroid U2 -> Now a U3 - This is a great board (quad core 1.7ghz), but higher cost ($65, plus more if you use emmc). They have a unique way to connect emmc. They have a connector that plugs it into the board. They also ship it with an adapter that allows you to connect it like a microSD card and update directly on your PC. I have 16gb versions of these. I like the new board as this one has some expansion capabilities (8 pin port that has UART/I2C) and they have expansion board that I2C gives you 16 IO pins, and UART is connected to Atmega328P and 20 IO pins…
Will be spending a lot of time with this board! The U2 was nice but I thought it would be hard to connect to robot. Note: I know there others using the Odroid Xu on their robot. This is an 8 core with lots of IO, but it is also bigger.
Intel NUC - Nice high powered intel box with an I5 and a 120gb emmc hard drive - Nice for power, but no IO. But again depending on what you want, you can plug in lots of stuff to the USB ports.
I have my code base setup to run on all of these, so I can continue to experiment.
I don’t think Lynxmotion needs to go this direction right now. However, you should also should keep a very close eye on boards like the Arduino Tre and UDOO, that combine what I believe is the best of both worlds. I’ll be getting in line to get an Arduino Tre when they are released, and I will definitely be getting an UDOO quad soon.
I think the Arduinos, including the Mega 2560 and Due have a lot of good to offer right now. As you probably know, I am in the process of migrating my code for W.A.L.T.E.R. from a regular Arduino (BotBoarduino) to an Arduino Mega ADK board with a DF Robotics Sensor Shield. I think Lynxmotion would do good to go with a tiered controller line up - from the BotBoarduino, to a Mega 2560 based board, and on to a Due based board. Another great option would be to create shields for existing boards - and this may even be the best way to go. There are a lot of excellent controller boards out there, with the capability of having add on boards created for them, but as is they just aren’t suitable for robotics. After seeing how well Lynxmotion has done with the BotBoarduino, I think you could really do well making robotics oriented shields for existing controllers.
While I really like the possibility of using a smartphone with robotics, I really don’t care for having them as main controllers. As for myself, I see a smartphone or tablet being used as an excellent remote controller. Take a good look at what Parrot has done, and you will see an excellent example of what I am talking about.
I also have to say though, the reason I got the Arduino Mega 2560 ADK board, instead of a regular Mega 2560, is the possibility of creating accessories that can be used with a smartphone or tablet (it doesn’t matter if it’s iOS or Android, but I prefer Android). I can see a smartphone app being used with a hardware accessory that has specialized hardware to remote control a robot. I think this is a very interesting possibility that should definitely be explored.
It’s not so much what I want the main controller on my robot to do, it’s that I have the hardware and software capability available to control my robot the way I want it to work. You make excellent hardware I can use to control robots, and I’ll take care of the software. Oh, of course you can always feel free to create libraries and tutorials for your hardware, just like Lynxmotion has always done.
Like I said above, pay very close attention to boards like the Arduino Tre, UDOO, and similar boards. I strongly believe that boards that combine the best of Linux and Arduino or similar platforms are the wave of the future for robotics controllers - IF they get the support they need to move them into robotics space (shields, other add ons, etc.). This is the kind of area where Lynxmotion can really shine!
Hey, here’s an idea for Lynxmotion: Make an interface that makes it possible to use LEGO Mindstorms stuff with SES brackets, BotBoarduinos, etc. This has already been done to some extent with shields being sold by RobotShop now, that interface Arduino and Raspberry Pi with LEGO Mindstorms NXT sets. This is another way to bring robotics to the masses, and I am planning to get at least one Mindstorms set for myself. I have two Raspberry Pi 512 Mb boards now.
I think Lynxmotion can cover a wide range of robot builders very well. They already do great with beginners to intermediate, on to really advanced builders. I think I probably fall somewhere in the middle part of the intermediate range, and people like you, Xan, Zenta, etc. are definitely at the high end of the very advanced builder range.
I do think Lynxmotion needs to embrace other types of effectors, including other types of servos like Robotis and such. I don’t think it would be that difficult to design and manufacture brackets that could bring the SES, Robotis, and other robot building systems together well. I’ve created some of the 3D CAD models Lynxmotion makes available, and I’ve looked at the possibility of designing hybrid brackets. I don’t have any capability to actually make what I design though.
At some point, I am going to have to bite the bullet and get another computer that can run Windows, because my Alibre Design Expert software requires it. I need to get back on maintenance ($499.00/year), so I can get up to the current version. If I could just pair this software with a nice 3D printer, I would have pretty much everything I would need to make custom stuff for my own robots.
At some point, Lynxmotion is going to have to start designing robot chassis that have places to mount larger form factor controllers as well as the smaller controllers like the BotBoarduino. It might also be possible to adjust the designs of their current robot chassis to fit both well. I am strongly in favor of seeing Lynxmotion go with starting to offer shields and add ons for existing controller boards. The “tack on” approach has never really appealed to me, and it often just doesn’t look or work quite right. A solution needs to be designed that fits with the whole robot.
I think the various Arduino board form factors are good sizes to be able to use, and a common mounting scheme (maybe using adapters) could work. There could, for example, be a larger opening for an Arduino Mega or Due board, but there could also be an adapter panel to fit over that which would fit a BotBoarduino and/or SSC-32 (separately, or as a combination). Adapters could be made that fit down, within a larger existing space.
To me, it doesn’t really matter what platform “comes out on top.” I am going to use whatever platform I can get easy access to (easily afford) that will do what I want and need it to do. I pick hardware according to features it has that I want and need to use. If I can make a way to afford that, I get it and use it. I still think combination boards like the Arduino Tre and UDOO are the wave of the future, because they are designed to have all the parts work together (or separate also, as in the UDOO).
If you haven’t at least looked at the UDOO, you really should, and I think you will see what I mean. I’m not saying that the UDOO is the final solution, but that a board of a similar design might well be. How about a board like the UDOO where you can use both of the main parts separately or coupled together as a single unit? A person could use just the features they need now (like, say, Arduino type features), and grow into and learn about the rest when they are ready (coupling the Arduino with a Linux system). Maybe the board is even modular, so a person doesn’t have to buy more than what they need now - Arduino and Linux based modules that can be attached to a base board as needed. This would take some very careful design work to have it work well though.
I would LOVE to have a platform I can install and run ROS on. However, most of those are in the $2000.00 plus cost range, which just leaves me out. I know there are various instructions for installing and using ROS on the PandaBoard (I have an ES), BeagleBoard, BeagleBone, etc, but I have never been able to get any of that to work. There is always something broken in the procedure, it’s way outdated, there is something not done, something is missing from a repo, or it just doesn’t work.
I am on a pretty low fixed income, and there is no way I am ever going to get my hands on a platform that’s that expensive. It’s taken me almost seven years to get to a point where I can actually bring W.A.L.T.E.R. 2.0 to an operational condition, and this is just ONE robot. I really envy those of you who have even just two robots to tinker with.
Yes, indeed! You can no longer purchase a controller board based on price alone. You have to look deeper in the provided functions, software availability, support, community, etc. In short, you have to learn more, or already know more, in order to make a reasonable decision. You also can’t just dismiss a particular board, just because of its low price. I almost made this mistake with the Raspberry Pi, but am very glad I took a second look at it.
It’s good to see Lynxmotion is going to branch out to support other technology.
This is exactly my thinking!
Yes, yes, and YES! There is no such thing as a board with too much memory, too fast a CPU, or too many cores… It should have 2 Gb minimum RAM, SD card capacity up to 64 Gb or whatever the largest size at the time is.
Not necessary, at least for myself.
Bluetooth and WiFi should be standard. HDMI is growing to be a requirement for development and testing, especially for boards that can run full operating systems with GUIs. At least two onboard USB 3.0 Host ports, plus USB OTG.
No. This depends too much on the design and capabilities wanted for a given robot. For example, I may favor different components for a similar function than Kurte or others would favor. Sensor choice can be a very individual thing.
I personally like this kind of scheme.
This is the appropriate and best way to go. Setup an image you know works with your hardware and product line. Then a person can customize that as required. It HAS to be easily customizable, mean it has to start out based on a standard distribution of an OS.
You REALLY need to, and should, take a close look at the UDOO… No more, will I say.
I’ve had two BeagleBone Blacks, which I really like a lot, as far as features. It’s much too difficult to get some things, like I2C and SPI, working though. These things should be working out of the gate. Yes, the information on how to do all this is on the interernet, but how does one tell which is the correct information from what is incorrect if they don’t know already? I did get I2C working, and it isn’t as difficult as it looked when I was trying to figure it out. I killed one BBBlack, and the last one just stopped working for unknown reasons - it was fine when I went to bed, but was dead the next morning. I can’t emphasize how much I really like the BeagleBoard Black enough, so am very anxious to see how the Arduino Tre works when it comes out.
Hmmm, OK, I am going to have to take another look at the U3 then. I don’t think it is going to trump an UDOO quad for me though. If there is an I2C expansion board, it better have a lot more than just 16 more I/O lines for me to be interested. More analog inputs are always nice, for instance. More UARTs are also always nice, and almost always in short supply.
I am interested in the XU, but don’t know if they are available here in the USA. There is no way I am paying a $50.00 shipping fee to buy one direct!
I have looked at the UDOO and it is a nice board. If I needed that level of IO connections, I might consider one. Right now, I will hold off. I have enough Linux boxes to keep me busy for awhile
Beagle Bone Black - Wonder what happened to yours. As I mentioned in other posts, there have been some who have resurrected their boards by playing with the power and the reset buttons… Also not sure if you are booting off of EMMC or sdcard. If SDCard maybe check your sd card did not go bad. Do you have the debug cable and does it show anything? When I was using angstrom Linux and had times where if I did the sudo apt-get update (or the equivalent on Angstrom) the board would no longer boot.
I am finding it works a lot better with the Debian builds. You might try downloading one of the new test builds and see if it boots. Also they are working to simplify the configuration of IO pins and the like. You should look for the builds by Robert C Nelson. up on the BeagleBoard.org. The one I downloaded and mainly use is off of: elinux.org/BeagleBoardDebian I believe he is also working to have the defaults configured to automatically support the prototype board, which I don’t remember what all they were bring out on this cape… I have downloaded the test build from Friday but have not fully tried it out yet.
Again waiting to get my hands on the Tre!
Then there is going the opposite direction and try to build a stripped down linux board. Today received an email on LinuxGizmos for a kickstarter for a linux board for robots… linuxgizmos.com/tiny-hacker-sbc- … ux-distro/
As for the cost of shipping of the ODroid, yep $40 is pretty high. Lots of email going on about this up on their forum. Up till now, they have resisted having distributors, but this last week or so, they have been looking at having some one over here to maybe ship stuff. So maybe soon.
As for using Teensy. Here is project from some other users playing with Teensy: dhylands.blogspot.ca/ May have to build a board that does what their breadboard does, plus a few more things I am interested in it having.
It’s not so much the added I/O connections I am interested in on the UDOO, but the integration level between the Arduino Due and Linux sides of the board that are possible. As you may remember, I have been saying for years that this is the way things should go to make Linux viable in the embedded arena.
I always boot off an SD card. I won’t touch Angstrom with a 10 foot pole. I’ve been running the latest Ubuntu release, and just started tinkering with the latest Debian release. The board is just dead to the world - nothing on the debug cable at all. I haven’t been motivated to fiddle with the buttons to see if that has any positive effect on booting.
Interesting. I just happen to have configured SD cards for the BeagleBone Black with Debian 7.2 and 7.3, so I will give them a try. I am really getting tired of Ubuntu.
Me too, since I took a close look at it! I will get in line for one!
Rex does indeed look interesting. I just looked at their Kickstarter page, and they don’t have any rewards between $120.00 and $2000.00. It’s no wonder funding is going slow for them now. Also, this statement really bothers me: “A standard Linux filesystem which will allow you to install just about any Linux software that can be cross-compiled for ARM.” I added the italics emphasis. Their statement indicates (to me) that users are going to potentially be left to build their own software by cross-compiling. Have these folks not learned anything from the likes of RedHat/Centos (RPM based) and Debian/Ubuntu (DEB based) about software repositories? Of course, their statement could also mean any software that can be cross compiled for ARM will be available from repositories. It’s ambiguous and needs to be clarified.
I’ve done a lot of cross-compiling. It’s is not a pretty sight, even when it does work right.
I just looked at the information on this and microPython. O M G ! If microPython can run on a Teensy 3.1 as Dave Hylands (one of my favorite hardware people) indicates, I MUST HAVE THIS NOW. With the right carrier board for a Teensy 3.x, this could be MAJOR.
Right now, I have just over 2600 lines of PYTHON code for my Device Control System (that’s just the main library), and it’s no where near complete yet. W.A.L.T.E.R. 2.0 is going to be at the center of the DCS - I’ve already proven it works, including authentication between devices belonging to different networks.
I’m rewriting everything in Go though. You can find a fork of the whee/adn Go package, I have heavily modified and added to, as well as the public Python sources for the DCS online. I’ll be basing my Go rewrite of the DCS on the Go package when it has all the functionality I need and works as I want it to. I have a private repo for development of the DCS.
As I mentioned before I have been having some fun playing with a Teensy 3.1 and wondered what it might look like if I made a breakout board the size of the botboarduino. This is not done yet and not sure how far I will take it, but here is a WIP picture showing a possible layout…
I still need to clean up the design, plus if I continue far with this may change any small surface mount components to something bigger and easier to solder. Probably won’t go all through ole but something bigger than 0805 components.
Still playing with functionality, like currently it shows
XBee (Default jumpers for Serial2), Speaker, 3 AX type servo connections with circuit protection (may relax this some), also will make it like some of the other parts where you remove a jumper and no longer tied to Serial1…
The bottom part has normal digital pins, where I currently have you can jumper groups of 8 to +5v or VIN. Top groups are Analog pins, currently all connected with 3.3v. May add selection for one set. To right side of Teensy is another 8 pins without power, that I may optionally add, but these require connections to pins on bottom of teensy… Have Voltage Divider to Vin by default goes to A0.
Toward bottom right is 3 pin connections for the serial ports, again Serial1 (if you use the Jumper (not in yet) to D1 defaults to Dynamixel servos, Serial2 defaults to XBee and Serial3 is free (Dups of D7, D8) Might drop these if I need more room for other functionality.
Could add PU resistors for SCL/SDA if I think I need it… Could simply put in traces and space for them…
However how many of us know enough to reconfigure the Beagle Bone and/or RPI to remove everything not absolutely required to get it up and booted? Note: there are those up on the BeagleBone forums, who think all of us who use them, should fully understand how to build and update the kernel and if you run into problems, you should be able to diagnose and build your own patches to make it work…
