Announcement

Collapse
No announcement yet.

First please abandon this JUNK hardware!!!!!!!!!!

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • First please abandon this JUNK hardware!!!!!!!!!!

    This is my 4th year of mentoring kids. I love teaching them java. I love working with them on the robot. But I absolutely can't take this hardware anymore. This will be my last year of participating in FIRST. All of the hardwork torpedoed by this terrible terrible terrible hardware. That is all.

  • #2
    I believe I found my issue. When I flashed to version 1.8.2 I lost the ability, on two robots, to scan the bus reliably. I went back to 1.6.0 and then I could scan the bus reliably but it still didn't work. I realized that I had to redo my configurations. My battery died before I could test it thoroughly.

    I get first has put a lot of work into this hw and the sw side of it is great. Please abandon the android phone and their flaky connections. Go to a system where everything needed is embedded in the controller. I love doing this but every year I have to sacrifice time with my family, job and sleep because I am fighting these connection and hardware issues. Then the kids go crazy because they are board because the mentors are trouble shooting the hardware.

    It is not fair you guys give us this inferior setup. This forum is filled with nothing but I can't see module 2, my phones won't connect together and many others.

    Comment


    • #3
      I respectfully disagree with your assessment of the technology issues. First has done a very admirable job of putting together an ecosystem that works for the most part. For those of us that work in tech fields, there are always constraints between development investment vs choosing available tech, which is the challenge here. And in the real world, hardware and software never work as planned, especially as the complexity increases. Fortunately there are companies like Rev that have created pieces of the solution to enable First to put this together.

      I mentor a team for a number of reasons, the chief one is that not enough kids are choosing a career in the sciences. I spend as much (sometimes more) time in teaching kids the application of how the technology works, and try to make the points as interesting as possible. This is all about getting kids excited about tech, learning as much as possible in a fun way.

      I am fairly new to First, so I don't know what other resources are available, but one recommendation I would have is some resources that help teachers and mentors bridge the gap between the First ecosystem and how things might be implemented in industry. For example, most microcontrollers run in a C environment, and the long unpredictable loop times of the First Java implementation wouldn't be acceptable for control applications. I teach this in a positive way, though, to the teams. It becomes an unavoidable constraint for the teams, it is what you have available given the budget you have, so find solutions that make the best of what you have.

      Comment


      • #4
        I agree with 11243_Mentor. In the four years we've had the new control system, we've gone from complete totally broken garbage in RES-Q (no documentation initially & internals distributed as binary-only, USB disconnect == immediate crash, "MR Motor Controller is returning Garbage Data" errors every other time the program was run, having to power off for 30 seconds to clear errors, etc.) to something that actually works pretty darn well (especially when paired with REV Expansion Hubs) given the constraints that it's still based on the phone.

        That being said, there's still a LOT more that could be done to make the system better, butnow that things are actually pretty stable, I don't think you could find a single person who would say the Samantha + NXT was better.

        Comment


        • #5
          I have to disagree with you guys. I have an ME and EE degree and I have been doing this four years. I am very familiar with how this system works and I help other schools with their issues. I don't see how a parent or teacher can get through these many many many issues. I agree the MR hardware was worse, but they are getting deeper and deeper into this flawed android phone system. From ESD issues latching the thing up to connection issues with the temperamental USB connections this is not a good system. A better system would have been to put the system on chip in the rev controller and have a USB camera. A raspberry pi, with a hat for the loads, would be better than this hardware. The issue is that so many of us have so many hours into this system its hard to admit the amount of hours we take away from our families, job and sleep overcoming the obstacles. I stand by my statement that this system is terrible.

          I am sorry to say it because I know many of you have put a lot of hours into making it. Probably a lot on your own time. I hate to be this blunt, but I genuinely can't take working with it anymore. I took 2 vacation days and got 5 hours of sleep all week trying to fix this bug for the competition. I have seen my family 1 hour a day.

          Comment


          • #6
            Originally posted by ejschuh View Post
            This is my 4th year of mentoring kids. I love teaching them java. I love working with them on the robot. But I absolutely can't take this hardware anymore. This will be my last year of participating in FIRST. All of the hardwork torpedoed by this terrible terrible terrible hardware. That is all.
            This is FIRST Tech Challenge. I can't understand why you would expect this program to be anything other than a Technology challenge. If you want a Robotics challenge you should switch to FIRST Robotics Challenge.

            All kidding aside, I have no problem with junk being produced in the first iteration of the engineering design process. More often than not, we've all conceived, designed, and implemented junk in our first iteration of our design projects. We then take what we've learned from the iteration and proceed to the next iteration of design or refinement. We exit the loop when an iteration of the re-design meets or exceeds requirements.

            The issue I would have is when you try to exit the loop early by changing the requirements to fit the design, instead of staying in the loop until a re-design meets or exceeds the requirements of a viable robotics control system.

            Clearly the control system has yet to fully meet requirements, but it has come a long way and has features far superior and robust than the NXT system. The control system will eventually meet requirements in virtue of the fact that, unlike the NXT system, the current SDK is portable to other hardware. If fact, there already exists readily available hardware today that likely meets or exceeds requirements. The SDK could easily be ported to better hardware.

            Big kudoos to the Tech Team for creating a portable SDK!!! In virtue of a portable SDK, market forces will force Rev and MR to continue to work towards meeting or exceeding requirements.

            Hang in there!

            Comment


            • #7
              Originally posted by ejschuh View Post
              A better system would have been to put the system on chip in the rev controller and have a USB camera.
              The REV Control Hub, already in use in FIRST Global, is on track to be approved for use next season (I think, anyway). The Control Hub is an Expansion Hub with a Qualcomm Dragonboard 410C built-in. The Dragonboard connects to the Lynx PCB directly internally with a pin header, and communicates with it over UART, effectively eliminating USB for anything other than a webcam which could plug into one of the two USB ports on it.

              Comment


              • #8
                Originally posted by ejschuh View Post
                I have to disagree with you guys. I have an ME and EE degree and I have been doing this four years. I am very familiar with how this system works and I help other schools with their issues. I don't see how a parent or teacher can get through these many many many issues. I agree the MR hardware was worse, but they are getting deeper and deeper into this flawed android phone system. From ESD issues latching the thing up to connection issues with the temperamental USB connections this is not a good system. A better system would have been to put the system on chip in the rev controller and have a USB camera. A raspberry pi, with a hat for the loads, would be better than this hardware. The issue is that so many of us have so many hours into this system its hard to admit the amount of hours we take away from our families, job and sleep overcoming the obstacles. I stand by my statement that this system is terrible.

                I am sorry to say it because I know many of you have put a lot of hours into making it. Probably a lot on your own time. I hate to be this blunt, but I genuinely can't take working with it anymore. I took 2 vacation days and got 5 hours of sleep all week trying to fix this bug for the competition. I have seen my family 1 hour a day.
                Regarding your comments about better solutions, who is going to provide this? Remember that First has limited resources, so their constraint is finding COTS solutions that they can adapt or marry in some fashion, and hopefully small companies like Rev take on the development for this market. And regarding your comment about leveraging a Raspberry Pi approach, this is a really bad idea due to the development and implementation cost required to create a full controller built around the Pi. I happen to know a great deal about this as I have just finished a new embedded design that would work in this type of application, and it is not as simple as you make it out to be. Raspberry Pis and their cousins have gone out of their way to make it diabolically hard to use them for embedded. Take a look at the attached photo. I am using an Orange Pi, set up as a wireless web server approach, where we use any phone as the GUI and control. I am using two separate processors interfaced to to the Pi, one that performs the control I/O functions, and one that simply manages power. Notice the supercaps and several switching supplies. This is required to safely use the Pi as power must be carefully managed or you kill the SD card. No app is required on the phone, it simply uses any browser.

                This is a niche market, so the development dollars and BOM cost of such a controller isn't something most companies would take on.



                Comment


                • #9
                  Originally posted by 11343_Mentor View Post

                  Regarding your comments about better solutions, who is going to provide this? Remember that First has limited resources...
                  Actually, FIRST has unlimited resources!!! FIRST has a vast army of volunteer electrical engineers and programmers at its disposal. This army of professionals would much rather dedicate their time and efforts on the challenge of coming up with better solutions vs. continuing to endure the types of challenges they've had to endure.

                  Comment


                  • #10
                    Originally posted by Alec View Post

                    Actually, FIRST has unlimited resources!!! FIRST has a vast army of volunteer electrical engineers and programmers at its disposal. This army of professionals would much rather dedicate their time and efforts on the challenge of coming up with better solutions vs. continuing to endure the types of challenges they've had to endure.
                    And yet FIRST continues to reject their help by doing things like distributing the SDK as "closed" source (binaries with source archives, which makes it impossible to make PRs).

                    Comment


                    • #11
                      You nailed it Alec. For example you can buy and off the shelf SOM from variscite for $40 -$100 dollars. This is cheaper than a phone. Make a simple expansion header on the REV hardware and you have a superior product than what you have today. https://www.variscite.com/

                      Comment


                      • #12
                        Originally posted by ejschuh View Post
                        You nailed it Alec. For example you can buy and off the shelf SOM from variscite for $40 -$100 dollars. This is cheaper than a phone. Make a simple expansion header on the REV hardware and you have a superior product than what you have today. https://www.variscite.com/
                        You are trivializing this, making me believe you have never developed anything similar to this that can be handed off to the legions of FTC teams with minimal integration knowlede. Why don't you volunteer to take on this project if you think it is as easy as you say?

                        Comment


                        • #13
                          Originally posted by 11343_Mentor View Post

                          You are trivializing this, making me believe you have never developed anything similar to this that can be handed off to the legions of FTC teams with minimal integration knowlede. Why don't you volunteer to take on this project if you think it is as easy as you say?
                          I didn't mention anything about developing new boards. I suggested to port the SDK to better hardware that is readily available today. But you have a great idea... A team can volunteer to port the SDK to better hardware and demo the hardware in action at the 2019 world championships. It would be very interesting project for a programming savvy bunch of students.

                          Comment


                          • #14
                            @11343, I have millions of controls in the field. If you own a Whirlpool appliance you probably use one of my designs daily. It is not that difficult to built a reliable system but the biggest mistake you can make is to not cut your losses and getting deeper and deeper into a system that doesn't work well. As far as volunteering I do that already and I have 100's of hours of helping other people fix bugs and root causing bugs on this system. I would rather have put that time into helping them with an architecture that works. Tossing a variscite SOM onto a daughter board that can drive all our loads and an LCD for configuration would not be that difficult. I bet a lot of companies would donate the resources to make it happen. For example Crank or QT would probably do the GUI in order to get their name out to hundreds of thousands of kids and mentors. A company like Jabil or Flextronics would probably design the hardware to do the same. Having your name as a supplier to FIRST is as big deal and that could be leveraged.

                            Comment


                            • #15
                              Originally posted by ejschuh View Post
                              @11343, I have millions of controls in the field. If you own a Whirlpool appliance you probably use one of my designs daily. It is not that difficult to built a reliable system but the biggest mistake you can make is to not cut your losses and getting deeper and deeper into a system that doesn't work well. As far as volunteering I do that already and I have 100's of hours of helping other people fix bugs and root causing bugs on this system. I would rather have put that time into helping them with an architecture that works. Tossing a variscite SOM onto a daughter board that can drive all our loads and an LCD for configuration would not be that difficult. I bet a lot of companies would donate the resources to make it happen. For example Crank or QT would probably do the GUI in order to get their name out to hundreds of thousands of kids and mentors. A company like Jabil or Flextronics would probably design the hardware to do the same. Having your name as a supplier to FIRST is as big deal and that could be leveraged.
                              Whirlpool appliances are like a lot of consumer embedded applications that do not rely on precise, real-time control. Your suggestion of using a SOM to control robots is a really, really bad idea, because they are generally architected with non-real time things in mind. A far superior approach would be to have a low-level microcontroller (code developed in C) that is doing all the robot control, with both loop testing and interrupt-driven control with the SOM. That way, things like PID could actually work reliably, accurately and consistently. I am having trouble understanding why virtually every discussion about PID motor control forgets about acceleration, because I have never seen it implemented, and it is clear that REV didn't implement it in the Run Using Encoder functions.

                              This also applies to sensing. The legal sensors for FTC have a great amount of noise and variability, like most sensors without some built in smoothing or integration function. What I would love to do is to take multiple sequential samples, throw out the outliers, then average them for the best result. This doesn't work in the current environment as the loop time can be anywhere from a few milliseconds to hundreds of milliseconds. This would a trivial function for a basic microcontroller.


                              Comment

                              Working...
                              X