Chief Delphi - [FTC]: Samantha? Like it or hate it?

Page 3 of 4

Show 100 post(s) from this thread on one page

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)

- FIRST Tech Challenge (http://www.chiefdelphi.com/forums/forumdisplay.php?f=146)

- - [FTC]: Samantha? Like it or hate it? (http://www.chiefdelphi.com/forums/showthread.php?t=87758)

normalmutant

19-12-2010 22:33

Re: [FTC]: Samantha? Like it or hate it?

Yesterday we had multiple disconnections during matches (even finals) at the NJ Frozen Frenzy Qualifier. This wouldn't be so much of a worry if it was just miscommunication between Samantha FCS people and the teams. That could be solved. But what usually happened was completely unexplained disconnections - sometimes making the NXT freeze up while driving and burn out the motors, sometimes just not running at all, sometimes shutting some teams off while other teams run and then turning them all back on. It just seems like Samantha is still in the testing phase. Or, at least, if they have figured out these problems, they haven't made it clear how to diagnose and fix them. I'm glad they're taking initiative - but I think switching over while Samantha still has some bugs is a bad idea.
Of course, I'm probably just bitter. I'd be a lot more open to it if it didn't cost my team 60 bucks every time we burn out our motors.

Random Dude

19-12-2010 23:25

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by PSHRobotics (Post 986641)

Are the FTA's allowed to change the initial configuration file that is put on the samantha during inspection? When my team was inspected (one of the first teams), the USB drive was taken out what looked like official packaging.

All of the FTAs have to setup their own router. Most will likely follow the official FIRST instructions, which use FTC_FIELD and FTC_PIT, (with a site chosen PW on FTC_FIELD). The USB drives used by the inspectors are created by the local FTA. I'm the head Oregon FTA, so thankfully I don't have to run any competitions until the end of January.

Quote:

Originally Posted by PSHRobotics (Post 986641)

See, this is partly due to a stupid bug in the FIRST teleop template (at least the RobotC one.) If communication is lost between to the brick, the robot keeps receiving the same joystick values it had been. Therefore, if you had been driving forward, your robot would keep driving until someone physically turns off the power. This can actually be tested under direct USB tether. Just drive the robot forward, and then yank the usb cable. I started coding a fix, but I need to send it off to FIRST so they can include it in next years competition. If someone knows if this applies to LabView as well, it would be appreciated.

Joachim

20-12-2010 09:47

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by wilsonmw04 (Post 986699)

I think we all need to take a breath here. We are talking about someone contentiously and maliciously cheating in a FIRST competition. This is not your average population we are talking about here. . . .

No it's not the average poulation. But the PA field operators at last year's PA championship event were apparently convinced they saw an attempt to do exactly that.

wilsonmw04

20-12-2010 10:03

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by Joachim (Post 986803)

No it's not the average poulation. But the PA field operators at last year's PA championship event were apparently convinced they saw an attempt to do exactly that.

I find that hard to believe.

PSHRobotics

20-12-2010 17:48

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by Random Dude (Post 986742)

Ah that sounds promising as long as more FTA's have the same foresight as you and switch to custom WiFi network names.

Quote:

Originally Posted by Random Dude (Post 986742)

This appears to be handled in Labview - I think all the templates that I have have this, because if an "Offline" message is sent to the NXT, the template is set to give the motors a power value of 0 and set the servo's to 0.
However, the problem normalmutant describes is with the NXT freezing after communication had been lost which would require the NXT's battery to be removed and re-attached for the NXT to be reset. (Or the reset button on the NXT to be pressed):

Quote:

Originally Posted by normalmutant (Post 986728)

But what usually happened was completely unexplained disconnections - sometimes making the NXT freeze up while driving and burn out the motors, sometimes just not running at all, sometimes shutting some teams off while other teams run and then turning them all back on.

My team and sister teams, luckily, did not have this problem during competition, but I do believe some of the disconnects did have this problem. I do know that this is not necessarily the Samantha's fault, because it did occur when using Bluetooth. I have seen it happen with the Samantha's during testing. I feel like its more likely a problem that the NXT has with disconnections. I do not know how to recreate the problem, but I have witnessed it with 3 different NXT's and 6 different laptops when using Bluetooth and 3 NXT's, 3 Samantha's, and 3 laptops when using Wifi (samanthas).

normalmutant

21-12-2010 12:41

Re: [FTC]: Samantha? Like it or hate it?

Sorry, I realize that my previous post wasn't accurate. A better description of it is that when the Samantha FCS stops working, it either shuts off completely, or continues sending the last signal it received from the joysticks. That's what I meant by "freezing up". This makes it harder, though, because the program can't tell when Samantha stops working - the signal it gets seems perfectly normal to the NXT.

Random Dude

21-12-2010 13:31

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by normalmutant (Post 987171)

It's actually not specifically a samantha issue. The same thing can (and does) happen over bluetooth, and even potentially direct USB tether (if one were to pull the cable out). The joystick.c template fails to get any packets from the controller, and just assumes the values should remain the same. (instead of entering a failsafe mode)

l0jec

21-12-2010 13:45

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by Random Dude (Post 987212)

If what normalmutant is saying about the Samantha locking up and continuing to send the same packet to the NXT is accurate, then a fix to the template may not work well without additional design changes? Your idea for a template fix to JoystickDriver.c as I understand it would rely on detecting that new packets are not being received so that you could stop sending the last known good one back in the joystick struct. If the Samantha cannot detect a drop and keeps sending the NXT old packets as if they are new ones, the template may not have enough info to easily know the difference without resorting to comparing the payload state of each packet to previous ones.
The "additional design changes" I mention above may be to add an incrementing id to each packet payload which could be used to easily identify if they are new or not by any downstream receiver (Samantha or ROBOTC/LabView code) and even allow tracking of dropped/missed messages.
Thoughts?

NalaTI

12-01-2011 08:05

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by Random Dude (Post 987212)

It doesn't need to be in the template. You can write your own and include it in your code. An example that could work is below.

Note that if the Samantha does keep sending the last packet, this won't work. But do we know for sure this is the case?

There's another problem I've seen occasionally that looks like a Samantha issue but isn't. If the I2C bus disconnects for a moment, it doesn't always re-attach correctly and thus the last message sent to the is continuously active. RobotC doesn't seem to have a means to verify and/or reset communication to the controllers programatically. That would be something that would REALLY help the issue.

Part of the complaining about Samantha is probably valid, but often it's the "newest thing" syndrome. When something doesn't work, the automatic blame goes to the newest thing. Even if the real issue has been there all along, or is something else.

Here's an idea for comm loss monitor code:

Code:

#include "JoystickDriver.c"



int nLastMessageCount = -1;

// This variable is used to keep a comparison copy of the message count



int nLastSysTime = 32767;

// this variable is used with the sSysTime system timer variable

// to create a pseudo up time counter.  Because it wraps automatically,

// that timer has an issue with a couple specific times. But since

// precision accuracy isn't required, it will still work for this purpose



bool CommOK()

{



    if (nLastSysTime < nSysTime) nLastSysTime = nSysTime;



    // The previous line handles the timer wraparound issue

    // in this one case, the delay before notification 

    // goes to a max of 3 Seconds if there's no comm.

    // I could have used a timer if we need the accuracy, but

    // there's only 4 of them and this way I don't use one up.



    // the variable ntotalMessageCount is defined and updated in

    // the joystickDriver.c file





    if (nLastMessageCount != ntotalMessageCount) // has there been a new comm?

    {

      nLastMessageCount = ntotalMessageCount; // update the comparison comm ID

      nLastSysTime = nSysTime;                 // update the comparison time



    } else {  // there has not been a new message



      if (nSysTime > nLastSysTime + 1500) return false;  

      // the previous line says that if our pseudo timer gets

      // to a second and a half, we've had no messages, and thus

      // comm is not OK.



    }

    return true; // if we get to here without returning false, comm is OK

}



task main()

{



  while(true)



  {



    while (!CommOK())

    {

      // Handle Comm (joystick message) failures here.

      // might be stuff like "motor[motorA] = 0;"

      wait1Msec(1);

    }

    wait1Msec(1);

  }



}

Shadowfl

14-01-2011 13:35

Re: [FTC]: Samantha? Like it or hate it?

I honestly can't say much about this question because it is my rookie year in FTC. but from what i know from technological evolution i conclude that the Samantha system does its job pretty well for its first year of mainstream FTC competitions.

aklego

06-02-2011 16:02

Re: [FTC]: Samantha? Like it or hate it?

I just FTA'ed at a small regional and we had two of the problems mentioned here: Signal drops and NXT lock ups.

Excepting the occasional loose wire, the signal drops were caused by low 12v batteries. However, I'm reluctant to blame that on the teams. The FCS uses essentially open circuit voltage to determine the state of charge. That is problematic, as a moderately discharged battery can still show green. Then, as soon as a load is applied, the voltage drops below the Samantha minimum and the module resets. We were fortunate in that our tournament had a 3-hr built in break. During this time, all the teams fully charged their batteries and the problem vanished. Teams that did all their practice and testing using tethering or bluetooth would never see this issue until their tournament. And teams that use many heavily loaded dc motors will be much more vulnerable to dropouts.

An inelegant fix for next season is to run samantha off of an independent battery.

NXT lock ups were the other issue. The only pattern I saw was that some robots never had the problem; the robots that did, however, had it repeatedly (but not consistently).

Random Dude

08-02-2011 17:28

Re: [FTC]: Samantha? Like it or hate it?

That pretty much mirrors what I've been seeing in Oregon in the 4 tournaments we've had so far. We've only had 1 NXT lock up, but Samantha losing power happened with some regularity. I definitely agree with the weak vs strong battery data.

As much as I don't like having another battery, I'm tempted to suggest to FIRST that Samantha have it's own. (I may also experiment with some of my local teams during the off season). The other thing I think will help would be splicing samantha directly into the battery connector, instead of way out at the end of the motor chain.

PhilBot

08-02-2011 21:46

Re: [FTC]: Samantha? Like it or hate it?

I wonder why they can't run the Samantha core from the NXT's battery.
Does anyone know if the NXT can provide power on the USB connector?

It Looks like Samantha sends the NXT battery voltage to the FCS. Is it reading it directly?

On a different track... it seems to me that the Sam is always showing a battery voltage that is low. I wonder if there is a series rectifying diode in the power feed. There's 0.6 - 1.0 volts that could be reclaimed :)

Quote:

Originally Posted by Random Dude (Post 1018619)

As much as I don't like having another battery, I'm tempted to suggest to FIRST that Samantha have it's own. (I may also experiment with some of my local teams during the off season). The other thing I think will help would be splicing samantha directly into the battery connector, instead of way out at the end of the motor chain.

aklego

09-02-2011 01:09

Re: [FTC]: Samantha? Like it or hate it?

One of the cool things about samantha is that teams can no longer goof and start the match with their power switch off. Powering samantha off a third battery could reintroduce that issue. That said, it would be pretty simple to put a relay in the circuit and have both the samantha battery and the main 12V battery controlled by the existing power switch. Or, perhaps even easier, replace the existing SPST switch with a DPST switch hooked to two batteries.

PhilBot

09-02-2011 09:34

Re: [FTC]: Samantha? Like it or hate it?

Quote:

Originally Posted by aklego (Post 1018944)

Although you can't start the match with the power switch off, there's still the problem if your Motor controller controller cables are unplugged.

Now that teams are relying on the Samantha module to report "no power", they sometimes forget to check the NXT screen for 12V levels. The NXT reads the 12V level through the motor controller, the Samantha does it via it's own power cable.

We had an instance where the FCS said everything was OK, but nothing moved. After the match we found that the NXT was reading 0V for the battery..... The motor controller cable had become unplugged. D'OH!

Phil.

All times are GMT -5. The time now is 20:49.

Page 3 of 4

Show 100 post(s) from this thread on one page