Our team opted not to go with a crab robot, but we noted the maneuverability advantage of a robot with such steering capabilites.
Has anyone decided to go with a crab-walking design (I know some teams have)? And if your team has, why did you choose that design over another option?
(There has been a lot of talk about crab-walking in the technical discussions, but I wanted to look at it strategically)
A good idea with the crab steering would be for king of the hill bots, they could strafe back and forth to block a robot from coming over (well, they could always just have wings, but if you didn’t)
Crab? OMNI DRIVE!
I beleive patrick posted something about how to pull it off, its really not that hard and allows you to go any direction instantly.
Greg
*Originally posted by Duke 13370 *
**A good idea with the crab steering would be for king of the hill bots, they could strafe back and forth to block a robot from coming over (well, they could always just have wings, but if you didn’t) **
The only problem is that most crab drives have high centers of gravity. A robot pushing up on an angle from below might be able to lift you up slightly and push you, or they’d tip you over altogether.
The only problem is that most crab drives have high centers of gravity. A robot pushing up on an angle from below might be able to lift you up slightly and push you, or they’d tip you over altogether.
I didn’t even think about that, but depending on how you design the rest of the robot, it could have a very low center of gravity. placing all the rest of the weight low (battery, frame, etc) could lower it drastically, or a complete redesign of the drive may solve the problem.
Well if your using omnidirectional wheels your robot won’t have much pushing power due to their design but with crab steering you should have an cool advantage because youll be able to move back and forth.
Why in the world would having an omnidirectional wheel affect torque? Yes, you’d be more easily pushed from the side if you had a pair of omni wheels, but your forwards movement is uninhibited… our strategy last year was to grab to goals and hold them their very well. We used omniwheels and could rarely be pushed in high gear. If we dropped down our second set of wheels (not omni), we could lift the goal which added more weight to the robot and more strength, thereby stopping us from getting moved. I think the only team that did so was 188 in Canada though we got them back in the finals 
An ideal omni-drive will cost you around 33% of your torque, but our robots are friction - not torque - limited.
Greg
Why in the world would having an omnidirectional wheel affect torque?
I didn’t say that did I. Omnidirectional wheels do allow you to get pushed from the side very easily due to the fact that you have put castors onto the side of it and not to mention omniwheels usually are made up of metal. This usually affects traction.
*Originally posted by monsieurcoffee *
**If we dropped down our second set of wheels (not omni), we could lift the goal which added more weight to the robot and more strength, thereby stopping us from getting moved. I think the only team that did so was 188 in Canada though we got them back in the finals**
Well, we pushed you, 340, and two goals across the field in NYC
But, no matter that, I really liked your method for changing wheels. It’s something we looked into this year, but ultimately decided against.
Our robot’s programming wasn’t quite working in NYC… lol. Stuff like the second set of wheels didn’t come down till about 3/4 of the way through the regional… stuff finally came together in Canada and also in the nationals.
190s wheels aren’t onmidirectional or crab steering, but we still won’t need to turn in most matches.
*Originally posted by wysiswyg *
**I didn’t say that did I. Omnidirectional wheels do allow you to get pushed from the side very easily due to the fact that you have put castors onto the side of it and not to mention omniwheels usually are made up of metal. This usually affects traction. **
I dont think you understand omni-drive
I’m talking about a robot with a 360 degree freedom of movement, instantly - no turning. From standing still it can go any direction it feels like.
And Ian:
I don’t rememebr this, and this is something I would remember
Sure, you pushed us accross the field, but we didn’t have 2 goals at the time.
Also, our method of switching gears was not weight efficent at all- the only reason we chose it was because it would let us lift up on the goal at the same time.
Greg
Originally posted by GregT *
**
Sure, you pushed us accross the field, but we didn’t have 2 goals at the time.*
Yes you did. 639 had grabbed both goals, and 340 was latched onto one of them also. [/nitpick]
warning of omniwheels: we had them last year and when we used last years bot on the ramp, we had to go up it backwards (omnis in back) because it would slide around…jsut a helpful tidbit
I think there is some confusion about drive systems… and locomotion in general.
On a plane (i.e. the surface of the playing field), describing an object’s position at any instant in time requires three coordinates. For example, a robot can have an x-location, y-location, and direction (angle) which the robot is pointing. You can not describe the robot’s position correctly with less than three coordinates. It is also possible to describe position with polar coordinates and other coordinate systems.
Now, over time, a robot can alter these coordinates. Typically, a robot can move forward and backwards. In other words, it can translate along one axis (move in the direction of the front of the robot). Most robots can also turn at the same time (adjust the angle which their robot is pointing). These two “degrees of freedom” are what you get out of a tank-drive system, which most teams choose to use. The number of degrees of freedom your robot has is defined as the number of coordinates (x-translation, y-translation, and z-rotation) that your robot can adjust simultaneously. A tank drive might be able to turn and translate in another direction, but it can not translate sideways, thus it does not have the third degree of freedom.
Typically, an omni-directional drive system is defined as a drive system with three degrees of freedom. Very few (I can only think of one last year) teams ever have three degrees of freedom. Tank drive only has two. In fact, even if you can turn all your wheels in any direction you like (i.e. swerve drive) you still have only two degrees of freedom, because at any instant in time your wheels are pointed in a given direction, and your robot is restricted to that linear and angular movement, giving you only two degrees of freedom. However, the advantage of the swerve is that you have the ability to change the direction of your prismatic (translational) degree of freedom with respect to your robot. If you can change wheel angles almost instantaneously, your robot is almost as good as one that can go accelerate in any direction at any angle, thus you virtually have three degrees of freedom. Robots that have a set of wheels that drop down perpendicular to your main set also only have two degrees of freedom, since at a given instant in time they can only move in one translational direction and rotate.
Now, a crab-walking robot could be built such that it has three degrees of freedom, but it would be difficult and almost certainly very very slow. The efficiency of an electric motor is far better than the efficiency of a crab-walking mechanism.
There only two mechanical ways I know of to get three degrees of freedom… meaning at any time, you can have any x-acceleration, any y-acceleration, and any angular acceleration. One of these I have posted a brief paper on how to get started on applying it to a FIRST robot (in the white papers) and the other is a little bit abstract and not too likely to work on a FIRST robot. One team had omnidirectional last year, and I forget the number, but I think it was a first or second year team. Basically it entails having three or four omniwheels perpendicular to the center of the robot. With three wheels, each unique combination of independent torques to the three wheels results in a unique direction and angular velocity of the robot.
One word: POWERED CASTERS
Sure, sounds impossible (or just very hard to do in FIRST) but someone I know came up with a design and an idea and I want to go and look at it.
From what my friend says, it has 8 options of movement.
*Originally posted by patrickrd *
**One team had omnidirectional last year, and I forget the number, but I think it was a first or second year team. Basically it entails having three or four omniwheels perpendicular to the center of the robot. With three wheels, each unique combination of independent torques to the three wheels results in a unique direction and angular velocity of the robot. **
You’re talking about team 857, they weren’t exactly rookies last year:
*Originally posted by Andy Baker *
**The core of this team has been in FIRST for about 7-8 years. These guys at Michigan Tech (Anthony, Justin, Adam, etc.) are some of the best robot builders in FIRST. They had one of the coolest drivetrains last year (the kiwi drive), and have been in the top 4 at the Championships before.We can all learn a thing or two from team 857. From what I heard, they really didn’t want to be rookies last year, but it was an administrative decision to do so.
Andy B. **
You can see their 2002 'bot on their site: http://stuweb.ee.mtu.edu/~alkrajew/FIRST/kiwi.mpg
*Originally posted by GregT *
**I’m talking about a robot with a 360 degree freedom of movement, instantly - no turning. From standing still it can go any direction it feels like.Greg **
This is also known as “holonomic” motion.
The long technical name we used for our ‘Kiwi’ drive system was an ‘omnidirectional holonomic platform with zero kinematic redundancy’.
Basically meaning it could drive in any direction (omnidirectional) like a swerve/crab drivetrain as well as do this with instant acceleration in any direction (holonomic) i.e. no need to turn the wheels like on a swerve/crab drive all while having no wheel drive in the same vector plane (zero kinematic redundancy) i.e. three wheels 120 degrees so no tow drive wheels produce motion in the same direction.
Ideally, the Kiwi (a true three degree of freedom drivetrain) can drive in any direction (x & y translational velocity) while rotating around its axis (rotational velocity). This is what a crab/swerve drive can not do. They can drive in any direction. And they can rotate on their axis. But they can’t do it at the same time. 
Now, what does this mean in competitions? 2002 was the perfect year for the Kiwi. We had three matched motors (the drills), a flat playing field, and a great deal of research ahead of time. A well constructed crab/swerve will give you nearly equal mobility with mechanically guaranteed straight-line motion (if done correctly, of course). Still, it won’t give you the fluid motion and overall ‘coolness’ of holonomic motion.
Adam
190s wheels aren’t onmidirectional or crab steering, but we still won’t need to turn in most matches.
Wait a minute, what could you do in this game with out turning – That’s crazy talk, man.