So normally this would be the cool part of the blog where I post up renders of my latest CAD work and concepts. Unfortunately, my CAD folder bit the dust at the same time as my old hard drive, so I’ll be analyzing the latest robot I made for a Swept Away scrimmage at the IRI.

I don’t have my design notebook, but with all designs I started with a thorough analysis of the game.

Anyhow, the rules are simple enough that I can summarize the main points here. You’re constrained to a small number of Vex pieces, and must start no higher than 12 inches. No other size requirements exist. You are allowed 5 motors and 1 servo. The point swings are 6 points for a football, 2 points for a softball under the goal, and 4 locked points for a softball in the high goal.

I initially began this as a collaboration with Nick Lawrence of 1503, so the original idea was that I would design the drivetrain and that he would design the manipulator. Unfortunately he enrolled in summer classes and was unable to help build or drive the final robot. I was able to borrow some parts from 1114 in order to finish the robot, though.

The first very big tradeoff was in drive gearing. With four vex motors in the drivetrain, and a fairly light robot, one could gear the drive for around 3.2 feet per second while preserving reasonable acceleration and drawing reasonable amounts of current. This gearing would give my robot a large speed advantage, mainly because it would take only a second and a half to go from wall-to-wall and 3 seconds to “switch sides” of the goals. The drawback is that only one motor would be available for the entire manipulator. An analysis of the game field showed a few “scoring locations”; the wall on the left, the wall on the right, and the goals in the middle. A mechanism that could only score footballs on the 12 inch sections of the field would need this speed to switch from “side to side” to avoid defense, but a goal scorer would not. So if goal scoring became the primary strategy, this could be ruled out almost immediately.

The next decision to make was what game pieces to score. Due to the very limited build time after Nick Lawrence backed out, it was determined that “doing everything” would be foolish, and priorities had to be made. The obvious decision here was footballs or locked softballs. I decided because the goals couldn’t hold more than 8 or so softballs, and that not having a football mechanism would put me in a hard to challenge points deficit off the bat, that footballs over the wall was the way to go. Because of this, the fast drivetrain option was still in play.

In the brainstorming phase, multiple complex mechanisms were thought up that could complete the game objectives. One early design called for a unique four bar linkage that tilted a spatula-like plate up to carry balls, and down to dump them over the wall. This linkage required two different crossed “bars” secured at different points on the robot, so it was deemed far too complex to be built in a day. A simpler mechanism used a four bar linkage with a roller claw to suck up and dump balls over the wall. The main concern with the design is that it would be very heavy, require many parts, and would require 2 motors, limiting the drivetrain.

I still needed that light and simple solution, that ideally used only one motor for the fast drivetrain. I finally finalized it on the plane to the event. The dimensions of the robot mandate a 12 inch maximum height, and a single jointed mechanism would get much higher than that away from the rotary joint (duh). But the arm would stay right at that 12 inch point, just with the manipulator raised in the air. The arm pointed up would allow things to roll down itself off the back of the robot, so a design that could “reverse dunk” over the wall would do just fine with only one motor. With a simple little deflector plate and footballs oriented in the right direction, game pieces could roll right off the arm and over the wall, with some velocity. So at this point I settled on an arm design with a passive scoop on the bottom to pick up footballs, that could “reverse dunk” over the wall. I was concerned that the robot wouldn’t be capable enough, but I didn’t have much of a choice so I reminded myself that all teams overestimate how much scoring happens in a game.

I got to the event and got to work on the robot. I didn’t finish in time for the first round of matches, which kind of sucked, but in the end it turned out to be just fine. For the first matches, my drive team (Basel A from 2337 and I as coach) decided just to “scrape” softballs in the holes in the field’s wall. This pushing of balls under actually scored a respectable amount of points and won one match, tying one, and losing the third. After that I finished the arm and ended up with this:

The vex robot between rounds 1 and 2

The “cradle” at the bottom held one football (not very well). Back up into the wall and drive the arm up and the football rolls down and over if you’re relatively straight into the wall and there aren’t balls underneath you. The arm was far from problem free, though. Most notably, because I did not program any trim into the motors, and because the arm was not balanced, the arm easily backdrove and could not hold itself in any stationary position. With a bit of latex and design work I could have balanced the arm but I had neither, so I made do by making the cradle more secure over time.

After two more rounds of qualifying, the robot was second and advanced to the finals, where it was promptly beaten by 1640. In both matches, the cradle bent completely out of shape by the end of the match. I don’t recall changing the design nor do I know why it suddenly started bending, but it did. 1640 scored more efficiently than my team did and was able to use their long arm to play defense despite our faster drive base. Overall, the robot performed very well for being built in a morning.

Here’s a Vimeo video I can’t embed for some reason.

If I were to redo the design with more time, I would only make a few changes. I would replace the collector made of bent single pieces with a design made with the 45 degree angle pieces and several standoffs, lengthwise. The football is a lot more rigidly held in between the gaps of said standoffs, and ramming the much more rigid design into a wall wouldn’t bend it. I would also add some surgical tubing to the back of the arm in order to attempt to “balance” it, fixing the rest of it with trim. With those changes the design would have scored faster than 1640 fixing the slow acquisition problem.

Advertisements