[JesseK]

2005: I'm a fan of 254's dual-PID arm with simple wrist joint. I didn't really get to see much that year though and that's only from perusing 254's old pictures.

2006: 25's turret, as seen in the behind the design book. The shooter wheels were suspended and allowed the balls to come up from a single point no matter what direction the turret was pointed.

2007 (overall): Gael Force (or Aces High??) for their inside roller claw. It had 2 wheels that spun outward and had high accuracy even with large driver errors when picking up an innertube. It was incredibly simple and very effective. Our Vegas/Atlanta claws were based off of it.

2007 (needs mentioning) 1731's 5-Degree Of Freedom (dof) arm WITH an identical arm used as its controller. This thing was a beast to see in person. The arm was a 2-dof wrist with 1-dof elbow and a 2-dof shoulder joint. The controller was the same thing except it was much smaller and used potentiometers to sense the desired position. The operator only had to move the control model in order to get the arm to respond. For complex movements it had about a 3-second delay, but for simple operations it was very quick.

2007 (drive train): 254/968's introduction of the ultra-light transmission with teflon-impregnated 7068 aluminum gears. 2-3 lbs (?) for a reliable shifting transmission is unheard of.

2008: 1565's "Twitch" drive which led to some pretty innovative control system techniques in 2009. Talk to 48, 1885, and 357 about for what happened in 2009 (though we all had varying levels of success due to implementation). It's essentially a swerve drive limited to two 90 degree states using omni wheels. The rotary action is pneumatic powered, and when all 4 links are independently controlled some whacky things can happen with the center of rotation.

2010: Nonadrive. Hands down, it's hard to believe there's a way to reinvent a skid steer to minimize all of the typical skid steer faults, but 148/217 did it. Granted it's a high maintenance drive train, but iterative design tends to work out all of the kinks. A team with mid-level skills and manufacturing abilities should be able to make a nonadrive from scratch assuming they understand the basics of traction and c.g. This single fact, imo, trumps 1625's complex 6-wheel crab for 'best' innovation in 2010.

I'm also personally a fan of MOE's tensioning and kicking mechanism in 2010 -- an effective motor winch system without the need of a dog gear.