Design flaw in BoeBot
Hi,
I was wondering how could the engineers of parallax have overseen such an obvious fault.
The ball on the rear (used as a 3rd wheel) rotates around an horizontal (while watching it from the rear) stable axis.So it has only got one degree of freedom.
As a result during the rotation of the whole boebot the ball slips and thus friction is developed between the ball and the floor,making it hard to rotate smoothly,putting more strain on the servos and making them wear out and consume more power.
This ball should have two degrees of freedom so as the boebot can rotate freely.The best in my opinion would be to use an omni-directional ball.
Maybe the results of the friction arent so noticeable but I still regard it as a serious design flaw,which moreover has a negative·aesthetic impact on the whole kit.
I was wondering how could the engineers of parallax have overseen such an obvious fault.
The ball on the rear (used as a 3rd wheel) rotates around an horizontal (while watching it from the rear) stable axis.So it has only got one degree of freedom.
As a result during the rotation of the whole boebot the ball slips and thus friction is developed between the ball and the floor,making it hard to rotate smoothly,putting more strain on the servos and making them wear out and consume more power.
This ball should have two degrees of freedom so as the boebot can rotate freely.The best in my opinion would be to use an omni-directional ball.
Maybe the results of the friction arent so noticeable but I still regard it as a serious design flaw,which moreover has a negative·aesthetic impact on the whole kit.
Comments
A "serious design flaw" is one that prevents functioning. A "minor design flaw" is one that impacts functioning. Perhaps using such an inexpensive method for a 'caster' does have an impact on power consumption. I'm sure if you came up with a solution that cost the same as a cotter-pin and a plastic ball, and that had less resistance in two planes of motion, Parallax would love to see it.
If it's really true that your perception of the rear ball "has a negative aesthetic impact on the whole kit", then it sounds like you're allowing one small issue to color your view of the entire solution. This kind of perfectionism is not rare in engineers, but it is something to be careful of. You need to be careful of it, because indulging this perfectionism leads to perfect optimization of one small aspect of your solution -- the rear pivot -- at the cost of the entire solution. Sure, a $10 omni-directional ball might be more "aesthetically pleasing" to you, but if it raises the price of the kit too much, nobody will buy it, or Parallax won't be able to afford to produce it.
There's lots of phrases to remind you of the danger of this approach. "Don't cut off your nose to spite your face." "Don't shoot yourself in the foot."
If it's still a problem for you, there are other companies that make $300 kits and above that use tank treads that don't use the "comprimise" of a plastic pivot ball. They have OTHER comprimises in terms of battery life, though. The over $1,000 market is quite sophisticated, as well.
Even if the cost of an omnidirectional ball is too much they could still add a vertical axis of rotation by separating a small part of the metal chasis with the ball on it and reattaching it on the same place on the rest of the chasis by using a single small bolt in the middle with only a few·threads at its end.This cant be so expensive,i suppose its a 3$ addition.I would do it myself if only I had the proper tools to cut such metal
However if the friction on the joint that i described,·is about the same as the friction on the ball then changing something is only a matter of aesthetics.So lab tests are required so as to determine the friction coefficients in each case.
Im not an engineer but unfortunately a perfectionist.
PS. Also I think parallax should release a better kind of rubber bands for the wheels so that they dont tear and thus it wouldnt be necessary to include 4 of them in the kit.
The ball looks to be made of UHMW polyethylene, a material nearly as slippery as Teflon. So the slipperiness is not confined to the outer surface, and it will never "wear off".
Engineering, as you must know, involves dealing with the tradeoffs extant in the real world. There is never a perfect solution to any design issue, and compromise — due to physics, cost, or both — must rule the day, or nothing will ever get done. Even the solutions you propose have their downside. Casters, because they're cantilevered, apply torque to their vertical axis, which translates to additional friction. And who hasn't fought a shopping cart whose rear casters malfunction? Dual axis balls have to be supported somehow, which by necessity translates to their outer surface. And that surface has to either roll or slide against its supporting structure. In either case, the amount of torque required to turn it will be greater than that required to turn a single-axis ball supported on a thin axle. Then, as if friction weren't enough of a bugaboo, there's "stiction", the tendency of surfaces at rest to adhere to each other until they break free and begin sliding. For a device as light-weight as the BOE-Bot, this could be the overriding concern. Imagine, if you will, a two-axis ball that won't turn at all because there's not enough force on it to break the stiction.
No, Parallax did the right thing with their rear ball. The tiny lateral axis requires minimal torque to rotate upon. And the single point of contact between the floor and a very slippery sphere makes turning on a dime effortless. What could be simpler, more cost-effective, and less prone to malfunction than that?
A design flaw? Hardly!
-Phil
Just my thoughts and reasons why I chose the Boe Bot as a starting point to get back into electronics and especially robotics.
Duffer
www3.towerhobbies.com/cgi-bin/wti0001p?I=LXRWW8
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST
1uffakind.com/robots/povBitMapBuilder.php
1uffakind.com/robots/resistorLadder.php
Please don't take something that was meant to introduce folks to robotics (or anything for that matter) and create a problem that the manufacturer has to FIX, thus raising the cost of the introductory kit.
As stiction I think you are refering to static friction.
A moving box is different than a moving wheel. In the case of a box before the motion starts static friction continuously increases until it reaches a peak called marginal static friction.Just after that the motion starts and we have·sliding friction which is slightly lower than the marginal friction and steady.So if the force is not enough to overcome stiction then we dont have motion.(i hope i have used correct english terms).In the case of a wheel when there is not enough force to break the stiction the wheel will roll,otherwise it would slide.When the rolling starts we have the rolling friction at the point of contact with the floor,which tends to slow down the wheel.This rolling friction is much less than sliding friction.Rolling friction is a necessary evil,and the goal is to eliminate sliding friction.If im mistaken somewhere please correct me.
Im not sure if I understand well but it depends about which axis are you referring to for the force to overcome the stiction,horizontal axis(ball-floor) or vertical axis(vertical metalic joint).On the horizontal axis we have rolling friction while in forward-backward motion,and i dont think we can ever have any sliding given the low speed of the bot.And overcoming stiction on the horizontal axis is not the goal for the above reason.On the vertical axis im not sure what happens(an engineer should answer this),if the hole at the joint is exactly the same size as the diameter of the bolt then we have sliding friction,but if the hole is slightly larger which is more probable then we might have both sliding or rolling,or only sliding or only rolling.Anyway it depends on lots of parameters and i think im already becoming annoying for you with my obsession on such a detail.
I agree with you Duffer,the boebot is·very good·in the electronic engineering part.
The sphere in the middle can rotate freely by any axis. Such a device in a smaller scale i think would be a perfect solution for the rear wheel.
If you PhiPi or anybody else is an engineer,please tell me how on the world this little device works(why the ball doesnt get out of the base)
I suspect that the base inside has got something like mini plastic rollers that form a circle a little bit above the equator of the sphere.
Thanks Zoot it looks really nice,perhaps ill go for it.
www.pololu.com/catalog/category/45
Nice when space is tight and you know surfaces will be relatively smooth (e.g. wood floor, table top, etc). I like the tailwheels myself because they provide a little bit of "spring" and roll over rougher surfaces pretty well.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST
1uffakind.com/robots/povBitMapBuilder.php
1uffakind.com/robots/resistorLadder.php
This is exactly what I had in mind as an omni-directional ball(different than the two-axis tail-wheel)! The cheapest costs 3$ and the most expensive 7$. After all Parallax could easily adopt one with minimal increase in price.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Engineering
-Phil
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Still some PropSTICK Kit bare PCBs left!
Real rubber wheel, ball bearing casters (as some people use on larger robots) will give the lowest rolling resistance in any direction, but they actually cause directional instability when changing directions, since the wheel carrier has to pivot off-center before it rolls. A real caster wheel in a useable size for the BoE-Bot is hard to find. Here are some doll-furniture casters, but I doubt they're any good. http://www.miniatures.com/hbs/global/Index2.asp?T=shopping&S=products&I=aa_ProductDetail.asp&P=1104·
BTW,·consider the CBA robot, which has a simple drawer pull as a skid. As long as there's not much weight on it, and the rubber tires have good grip, that's just about as good as one of these ball casters. http://www.budgetbot.com/CBA_Kit.html
It's hard to second-guess those Parallax guys. They're sharp.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·"If you build it, they will come."
However, if you want a nice friction-free tail wheel here's what you can now buy from Parallax at $79.99. It has dual sealed bearings, 20+ lb weight capacity and it rolls sideways. This tail wheel goes with our 12V Motor Mount and Wheel kit. However, it was designed without any load-bearing, friction, or pressure calculations in mind. We·have 100 units in kitting right now.
Ken Gracey
Parallax, Inc.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Whit+
"We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
There are two kinds of friction: static and kinetic.
Static friction is what prevents stuff from sliding around (like a book an an incline, a nut on a screw, or a car parked on a hill)
Kinetic friction is part of the effort required to move stuff around (every try dragging a box?)
Now onto wheels. The friction between the outer rim of the wheel and the ground is what turn the wheel. In most cases (drag racers excluded) the outer rim of the wheel does not move relative to the ground when it is in contact. Therefore, it has static friction. That sounds bad, you say. Yet this translates into mechanical advantage, and so the friction that you want to consider is that of the axel and the bearings. The wheel is not a wheel unless there is static friction on the outer rim and kinetic friction somewhere on the inside.
To translate to engineering: you can get perfect wheels if you eliminate kinetic friction from your designs. Since static friction does not take work to keep active, you won't waste energy to keep such a system in motion. Now, this is highly unlikely in the real world, but so far ball bearings can get pretty close. Think about the design: all the weight is on components that rely on static friction to operate. The kinetic friction is all on the sides, where there is much less Ffr.
Thanks for the information…if it was April 1st I would remind Ken that he forgot to mention our soon to be released friction-free anti-gravity drive system with ion propulsion.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Engineering
What an education to the novice robot maker !
Friction, degrees of freedom, materials science, economics lessons, trade-offs and payoffs, energy management, goal seeking, ..., ....
Brilliant choice !!
PAR
There's so much to robotics beyond programming. Any given aspect could be the focus of long-term experimentation. Some of my robots never even rolled - I used them to learn how to design a particular subsystem (like the engine/alternator/voltage regulator project).
Ken Gracey
Parallax, Inc.
Thanks for your info and suggestion erco. While the rubber wheels seem superior to those of the boe-bot,i dont like the rear "thing" even though you say its almost equal to the ball caster.Its just not aesthetically pleasing
Ken I suppose you are suggesting this tail wheel for a larger bot and not for the boe-bot since it looks huge.(is it an aircraft wheel?
SRLM thanks for contributing to the physics of the issue.What you call kinetic friction, I can deduct the conclusion that it is what I call sliding friction,maybe I havent used the best english term.I agree with what you say.You just forgot to mention the rolling friction because it also exists,even though that its not so popular.Mainly it has to do with the elasticity of the tire and the floor and their deformation during rolling which produces this kind of friction.But of course in the macroscopic world this is not even an issue.
Chris I demand that you release on your store the friction-free anti-gravity drive system with ion propulsion at the price of 50$
PAR· maybe· brilliant choice would also be if parallax released a unicycle boebot. You should see then how many threads there would be commenting on why doesnt it achieve dynamic balance.
stavros, whether or not the 12V Motor and Wheel Mounts are a fair price depends largely on something we rarely measure as hobbyists: our time! What if you had to build an R/C airplane from scratch? Would that experience make a $300 ARF kit more price-worthy? The same applies to robotics - we need all the help we can get with the time we have for this hobby. $300 spent on our 12V Motor and Wheel Mounts + a tail wheel gives you a tremendously capable platform that can carry a child. This is really hard to beat.
Considering the value of your time and the high quality of the particular product, I'd say it's more than a fair price. In fact, it violates some basic cost/pricing rules we've established by being a record low profit-maker coupled with high volume. These machined products exist only for our customers needs, and because they'll also sell our other hardware (micros, sensors, HB-25s, etc.) but on their own they're a tremendous looser. For you, they're an awesome value!
They're also made in the USA. One machinist hour with benefits costs us $80. The CNC milling machine is $55K and the engineer who designed the product spent three months of time with prototypes, electronics and documentation. We'd have to sell 10K units to start to recover our investment. . . but that's not why we made them. They're a real bonus for our customers. In any other company the accountants and marketing staff may squash such products. I just remembered that the motors are surplus - I had to buy $30K of them just to ensure a reasonable lifespan without redesign.
Competitors: don't do what we did! You might go out of business if you follow our open-sourced business model explained above!
The pictures on our web site simply don't do the 12V Motor Mount and Wheel kit justice. You'd need to see these up close to appreciate them. I bet then you'd say they're a reasonable price.
Ken Gracey
Parallax, Inc.
BTW, I'm enjoying the debate Stavros and hold no hard feelings ...
Edit: of course, the elasticity of the wheel wastes very little energy. You'd be better off making more aerodynamic than worrying about tire flex. (Perhaps 100x better off)
Post Edited (SRLM) : 9/26/2008 1:22:19 PM GMT
Ron
PAR
Plus, it's not a good idea to make conclusions like "I suppose people just don't like xyz...". When you do that, you stop listening to other opinions. That can be expensive.