<R08> Section M

[Daniel_LaFleur]

Originally Posted by Craig Roys

High speed in this game won't be more that 7 or 8 fps. You need traction to gain speed quickly. With 6 robots and trailers out there on a slippery surface, no one will be able to gain high speed for a collision. Collisions in the past 2 or 3 years were much higher. Also, it will take an awful lot of force to break 3/4" in plywood with padding in front - considering it will be hit with another padded bumper, I don't believe there is any way that a reasonbly supported bumper would break.

I am not anti-bumper here; in fact, I like being able to use bumpers. We've used bumpers every year since our rookie year in 2006; even when they were optional. They save a lot of robot wear-and-tear. What I do object to are bumper requirements that limit creativity with robot design due to the requirements. Take a look at the length of the bumper portion of the manual compared to other sections. In 2006 and 2007 the bumper requirements were two pool noodles covered by fabric, backed by 3/4" plywood securely attached to the robot. Last year they became a little more detailed due to the fact that everyone was required to use them. This year, they seem to have gone crazy with the bumper regulations.

Oh well, we'll deal with them whatever the rules are - some of the requirements just seem very unnecessary.

During 'Rack and roll' our robot, with a maximum speed of 6.5 ft/sec broke our front bumper (38" long, only supported on the ends) 3 times. I'll grant you that we played a lot of defense (who doesn't up here in New England) but we never received any penalties for ramming / playing too aggressive defense.

Having an unsupported bumper this year will invite broken bumpers.

[Cory]

Originally Posted by Daniel_LaFleur

During 'Rack and roll' our robot, with a maximum speed of 6.5 ft/sec broke our front bumper (38" long, only supported on the ends) 3 times. I'll grant you that we played a lot of defense (who doesn't up here in New England) but we never received any penalties for ramming / playing too aggressive defense.

Having an unsupported bumper this year will invite broken bumpers.

There's a difference between a 38" long bumper segment only supported at the ends and a 38" long bumper segment supported every 12". The latter is never going to break if you're using high quality plywood.

The example I gave is what I believe Craig is referring to.

[Daniel_LaFleur]

Originally Posted by Cory

There's a difference between a 38" long bumper segment only supported at the ends and a 38" long bumper segment supported every 12". The latter is never going to break if you're using high quality plywood.

The example I gave is what I believe Craig is referring to.

I believe that the impacts will be faster than 6.5 ft/sec this year as well, considering the standard gears given.

And not everyone is going to use "high quality" plywood.

[dlavery]

Originally Posted by Cory

There's a difference between a 38" long bumper segment only supported at the ends and a 38" long bumper segment supported every 12". The latter is never going to break if you're using high quality plywood.

The example I gave is what I believe Craig is referring to.

I hate to "break" it to you, but that is not quite the case. During early testing, I was driving a robot with bumpers attached in exactly the manner you describe (rugged stiff fasteners, robust stanchions about every 12 inches, bumper proud of the robot frame by about three inches). In the very first drive-from-one-end-of-the-field-to-the-other-as-fast-as-you-can test, the bumper broke and splintered upon impact with another robot at the far end. Within just a few minutes, I did it again while pushing the robot around manually, just to see how fast I could push it against wheel slip.

/edit/I just ran a few numbers out of curiosity. In a "perfect collision" situation (two full weight 151 pound robots hitting head-on at 9 fps, with one of the robots skewed so it impacts the other "corner first") the impact forces get pretty impressive. As the robots collide, they compress the pool noodles down to 20% of their original thickness in about 0.009259 seconds. At a closure velocity of 18 fps, this is a peak change in velocity of 1944 ft/sec/sec, or a 60.75-G impact. Since I said the robot impacted "corner first" I will posit an impact area of 1.5 square inches. Assuming the pool noodles absorb about 18% of the impact energy during compression (not too bad for material of this type), that still means that the localized impact pressure is right around 10,000 pounds per square inch. I haven't looked at the bending moment of 3/4-inch plywood on 12-inch support centers yet. But I am now really not surprised by what happened to the bumpers. /edit/

There is a reason for that rule. Don't count on it changing.

-dave




.

[Craig Roys]

I'm not counting on the rule changing and we'll just go back to our previous chassis design - we just thought it might be nice to try something new (or at least new to us). That's not a real big deal. I am a little suspect of being able to break 3/4" plywood supported every 12". If you point the grain correctly and you use "plywood" (as opposed to OSB or something similar) it should be near impossible to splinter. 1/2" plywood is commonly used on roofs with 24" inch centers between trusses - again, if the grain is pointed correctly it is very strong. Keep in mind that roofs in MI need to hold a lot of snow at times. I don't know what your set up was and it's quite possible that I'm wrong - it wouldn't be the first time. I may have to set up and run some tests just to satisfy my curiosity and skepticism.

But that's not my main argument here. Every team knows (or at least learns quickly - sometimes the hard way) that you need to build a robot robustly to compete in a FIRST competition; especially in elimination rounds. If a robot is not robust enough you will spend your time fixing it as opposed to competing. If a team tries to cut corners to save weight or $$$, they do so at there own risk knowing that the robot needs to hold up to the rigors of competition. I think that the same should go for bumpers - if you choose not to follow the GDC's recommendations for installation you do so at your own risk knowing that you can't compete with broken bumpers.

But, like I said; life goes on - we'll just adjust our designs accordingly and move on.

[eugenebrooks]

The language of the bumper rules in the 2009 manual
has been clear from the start, although there is no specific
specific specification for the thickness of the support
for the bumper, only that it must be there.

Additionally, the manual has indicated that collisions are expected in
the game, Bill has advised us to put the velcro on the bottom
of the driver stations in his blog, and now Dave has posted his
60 G estimate for worst case collisions.

I guess the word to the wise is to design the
electronics and battery mounting carefully.

Eugene

[dtengineering]

Originally Posted by dlavery

During early testing, I was driving a robot with bumpers attached in exactly the manner you describe (rugged stiff fasteners, robust stanchions about every 12 inches, bumper proud of the robot frame by about three inches). In the very first drive-from-one-end-of-the-field-to-the-other-as-fast-as-you-can test, the bumper broke and splintered upon impact with another robot at the far end. Within just a few minutes, I did it again while pushing the robot around manually, just to see how fast I could push it against wheel slip.

-dave


.

And that is why NASA has NEVER let Dave drive Spirit or Opportunity!

Nope... even better... "Well if you drive like THAT no WONDER your other car is on Mars."

Jason

[Gdeaver]

There are many grades and types of 3/4" plywood. From experience most plywood that teams buy at a typical home center is not the best for high impact and structural strength. A high grade Baltic or Finnish birch plywood would survive the impacts. How ever it is not normally locally available to all teams and is expensive. The average plywood teams use needs the reinforcement . I would agree with the GDC's assessment.

[artdutra04]

I decided to do some more math to see exactly how it would be possible to have robots carry out these super high speed colisions.

Using f=m•a and f=μ•f_n, where f_n = m•g on a level surface, solving into each other we get a=μ•g. Since gravity is 9.81 m/s², and the given coefficient μ of static friction in the game manual is 0.06, we get a maximum robot acceleration this year of 0.5886 m/s². It would take a robot ~4.66 seconds of maximum acceleration to reach 9ft/sec or ~2.74 m/s. During this time, the robots would each need ~6.39m to accelerate to this final velocity, or about 20ft. If their wheels were slipping the entire way, resolving for dynamic friction would give us about ~7.66m or ~25.1 ft covered while accelerating to top speed.

In order for this to happen, the two robots would have to be at opposite ends of the playing field, and simultaneously floor it and accelerate as fast as possible without wheel slip until then both hit each other head on. You know, kind of like what's going to happen in autonomous? ;-)

[Matt C]

Originally Posted by artdutra04

In order for this to happen, the two robots would have to be at opposite ends of the playing field, and simultaneously floor it and accelerate as fast as possible without wheel slip until then both hit each other head on. You know, kind of like what's going to happen in autonomous? ;-)

Are you implying that with the way the robots are positioned at the beginning of a match, that autonomous is going to be like some . . Robot Demolition Derby?

[Richard Wallace]

The 2003 game (Stack Attack) had four robots start the game by simulatenously charging up a ramp, trying to be first to hit a wall of bins and knock/plow as many as possible into their own scoring zones. The frequent result was high-speed collisions, mitigated (i.e., damped) in most cases by bins interposed between the colliding robots. Bumpers were not required back then so most robots didn't have them. Fortunately, many robots also lacked sufficiently powerful drivetrains to develop significant kinetic energy at the moment of impact; however, in a few cases the crashes were spectacular.

Lunacy will provide much more frequent crash opportunities. Bumpers designed to mitigate the effects of those crashes are not just a good idea, they are the law.

[Jon Stratis]

Originally Posted by Matt C

Are you implying that with the way the robots are positioned at the beginning of a match, that autonomous is going to be like some . . Robot Demolition Derby?

yeah... it should be fun to watch! Everyone's autonomous mode is going to be "get away from the guy right behind me chucking balls into my trailer"... and on top of that, everyone starts out pointed straight at the center point!

[johnr]

So the gdc probably isn't going to come out with a minimum requirement for bumper backing. So i say leave it up to the teams to decide how to protect their robot,but have a test at inspection. Maybe a 120 pound weight with a six inch bumper on it. Pull it back(to a set distance) and let it fly. If your bot 's bumper survives your good to go. If not ,at least your in the pits.

[dtengineering]

Originally Posted by dlavery

/edit/I just ran a few numbers out of curiosity. In a "perfect collision" situation (two full weight 151 pound robots hitting head-on at 9 fps, with one of the robots skewed so it impacts the other "corner first") the impact forces get pretty impressive. As the robots collide, they compress the pool noodles down to 20% of their original thickness in about 0.009259 seconds. At a closure velocity of 18 fps, this is a peak change in velocity of 1944 ft/sec/sec, or a 60.75-G impact. Since I said the robot impacted "corner first" I will posit an impact area of 1.5 square inches. Assuming the pool noodles absorb about 18% of the impact energy during compression (not too bad for material of this type), that still means that the localized impact pressure is right around 10,000 pounds per square inch. I haven't looked at the bending moment of 3/4-inch plywood on 12-inch support centers yet. But I am now really not surprised by what happened to the bumpers. /edit/

There is a reason for that rule. Don't count on it changing.

-dave




.

Okay... seeing as how my other car is still on this planet, I'm a bit hesitant to question these numbers, but I was doing some calculations with our programmers this evening to figure out peak velocities and such and have to question the assumed closure velocity cited here.

We used the published value of static coefficient (.06) of friction to determine that a 150 lb (68kg) robot would have a normal force of 668n and a peak forward force of 40N. The mass of the robot, plus trailer, is 186lb, or 84kg, giving a peak accelleration of 0.47 m/s/s

Next we assumed that the effective length of the playing field was 15m. Although 54 feet works out to be 16.5m, or thereabouts, the length of the robot and trailer, as well as the driver station bumpers must be subtracted from the space available for picking up speed.

Assuming constant acelleration, of .47m/s/s over 15m, it should take a minimum of 8 seconds to cross the playing field from one end to the other, with a peak impact velocity of 3.76 m/s or... 12.3 feet per second.

Now this is the peak velocity of a robot hitting the end... but it is also the maximum impact velocity that any two robots could sustain. If each started out at one end of the playing field, they would meet in the middle, and would each only have reached 6.15 fps each, for a closing velocity of 12.3 fps, which is just 2/3 of the assumed 18 fps velocity impact. (Actually it would be lower than 12.3fps, as the effective length of the playing field would again be diminished by the length of the second robot/trailer combo unit.)

That isn't to say that some robots might not exceed the published coefficient of friction as the playing field wears, or that a 12 fps impact is something to be laughed off without concern... we'll be building a solid robot and strapping solid bumpers on it... we agree with the point of the post and if this were anything but FRC would probably just say "close enough, good enough" on the calculations, but the peak closure speed and resulting extreme G-forces didn't mesh with our calculations and we were wondering if we had somehow missed something.

Or, perhaps, if the 18fps impact velocity is based on actual testing of robots on regolith, then the published coefficients of friction don't provide an accurate estimation of robot performance. I know a few teams have posted suggesting that their experimental results for coefficients of friction are much higher than the published values.

Any suggestions?

Jason

<Edit> first assumption... that is not quite right. We assumed all of the weight of the trailer would be over the trailer wheels. Some of it will contribute to the normal force of the robot and thus improve traction and accelleration. Even assuming 100% of the trailer weight does so, however, peak accelleration is just .6 m/s/s and it takes 7 seconds to make the trip with a peak velocity of 14 fps. We're getting closer...

second assumption... we were assuming a straight line path from one end to the other... it may be possible to achieve a slightly higher peak velocity by taking a curved path along the playing surface... </edit>

<edit 2> third assumption in these calculations is that accelleration will take place on the regolith. Maybe, just maybe, if everything is right and teams are driving at least partly on the carpet, an 18 fps impact speed is a theoretically possible event </edit>

[eugenebrooks]

Use the measured figures for the coefficient of friction and you will be closer to reality with your estimates. Several teams have posted measured figures on CD, and their measurements are roughly twice the inline published values.

I would also like to say that a driver who accelerates all the way down the field and then crashes into the back wall, or into another robot near the back wall, is not engaging in an accident. A reasonable expectation is that drivers will be required to attempt to maintain control of their robots, and will be expected to plan their acceleration and braking so that they arrive at their destination without a high speed crash. I would at least hope that this will be the case, although I have not yet run across this expectation spelled out as I read the rules. I will have to read the rules a little more closely, I guess...

Eugene

Originally Posted by dtengineering

Okay... seeing as how my other car is still on this planet, I'm a bit hesitant to question these numbers, but I was doing some calculations with our programmers this evening to figure out peak velocities and such and have to question the assumed closure velocity cited here.

We used the published value of static coefficient (.06) of friction to determine that a 150 lb (68kg) robot would have a normal force of 668n and a peak forward force of 40N. The mass of the robot, plus trailer, is 186lb, or 84kg, giving a peak accelleration of 0.47 m/s/s

Next we assumed that the effective length of the playing field was 15m. Although 54 feet works out to be 16.5m, or thereabouts, the length of the robot and trailer, as well as the driver station bumpers must be subtracted from the space available for picking up speed.

Assuming constant acelleration, of .47m/s/s over 15m, it should take a minimum of 8 seconds to cross the playing field from one end to the other, with a peak impact velocity of 3.76 m/s or... 12.3 feet per second.

Now this is the peak velocity of a robot hitting the end... but it is also the maximum impact velocity that any two robots could sustain. If each started out at one end of the playing field, they would meet in the middle, and would each only have reached 6.15 fps each, for a closing velocity of 12.3 fps, which is just 2/3 of the assumed 18 fps velocity impact. (Actually it would be lower than 12.3fps, as the effective length of the playing field would again be diminished by the length of the second robot/trailer combo unit.)

That isn't to say that some robots might not exceed the published coefficient of friction as the playing field wears, or that a 12 fps impact is something to be laughed off without concern... we'll be building a solid robot and strapping solid bumpers on it... but the peak closure speed and resulting extreme G-forces didn't mesh with our calculations and we were wondering if we had somehow missed something.

Or, perhaps, if the 18fps impact velocity is based on actual testing of robots on regolith, then the published coefficients of friction don't provide an accurate estimation of robot performance. I know a few teams have posted suggesting that their experimental results for coefficients of friction are much higher than the published values.

Any suggestions?

Jason

<Edit> first assumption... that is not quite right. We assumed all of the weight of the trailer would be over the trailer wheels. Some of it will contribute to the normal force of the robot and thus improve traction and accelleration. Even assuming 100% of the trailer weight does so, however, peak accelleration is just .6 m/s/s and it takes 7 seconds to make the trip with a peak velocity of 14 fps. We're getting closer...

second assumption... we were assuming a straight line path from one end to the other... it may be possible to achieve a slightly higher peak velocity by taking a curved path along the playing surface... </edit>