First cuts are made! A task that every CyberKnight will learn and participate in every year is the joy of deburr-ing and edge filing. The robot is actually, physically, starting to take shape.
Seeming Simple Is Not Simple
At the end of kick-off, myself, several other mentors and several veteran students all had built up the same false concept in their heads: wow, this looks pretty simple. I mean, you've got a disc, and you've got a ball. We have so many sources for in-taking, storing, and shooting/expelling balls. Discs, a little different, but still easy. I mean, they are giving us these Hatches from a known fixed position. They have hook-and-loop on them. They re-attach with more hook-and-loop. Place a hatch here, shoot a ball there, do it a few times more than the other team, and you win! How hard can that be?We've had 2 full weeks to break down nuances, strategize, play-act the game, watch MCCC, talk to other teams, and more. Myself, other mentors, and several veteran students have come to the same conclusion: this is not a simple game. The mechanisms to efficiently obtain and deploy game elements, factoring in defense, thinking about optimal robot paths, game element scoring sequences, and well, we've already consumed 2 weeks of build.
Cargo: These playground balls are not given a specific size or inflation metric, they bounce erratically, they wind their way to inconvenient locations. Collecting cargo requires teams to build intakes or holders that can manage multiple inch deltas among game element diameters. if teams collect over the bumper, they will need to think about robust mechanisms taking a hit from other robots, or collisions with walls and the field. Teams collecting within their frame perimeter will have to think hard about making collection fast and easy for their drivers. Scoring cargo requires being able to raise the cargo to four different heights, rocket levels 1, 2, and 3, and scoring cargo in the cargo ship may be yet another. In general, cargo requires hatch panels get placed before they can/should even be thought about and collected!
Hatch Panels: They have a hole in the center through which they can be grabbed. Teams can grab the outer rim of the hatch panels. Teams can use the hook-and-loop to collect hatches. Can your robot hold the hatch panel securely? Dimensions plagued our decision making when looking at the hatch panels. There is a 3-inch gap behind the hatch panel at the load station, and a 3-inch gap behind the panel scoring locations in the rocket and cargo ship. When scoring the rocket however, there is an additional ~3-inch distance that the robot will have to overcome, as there is no bumper slot cutout. This potentially adds another mechanism. This may add weight, system complexity, and operator complexity. This also now means a mechanism will extend beyond the frame perimeter. Mechanisms that extend beyond the frame perimeter must be made robust, able to take a hit from field elements. If a robot is defending you as you try to score, the mechanism may take lateral loads when lining up, or attempting to deploy. A broken mechanism can't score points, so how do you balance out mechanism speed and lightness, alongside robustness and resilience?
The field: Loading zones are in the corners. Hatch panels are held by door brushes, which take an amount of force to pull through. Or take extra time to lift up and over. Cargo balls will be strewn across the field. They will get into the loading zone corners and will need to get pushed or rolled away. Cargo can get caught under taller bumpers and hang up robots for precious seconds as they attempt to un-stick. The space between the rocket and the cargo ship is huge! Except that it isn't. Two 30" or larger robots will struggle to cleanly navigate past each other in the space between the rocket and the cargo ship. The HAB level 1 is a raised platform, it can catch an un-prepared robot chassis, and it can throw off robots running automated paths.
All of the above, and more (climbing the HAB, defense, strategic scoring hatches and cargo for maximum points and opportunities) go into each of our mechanism reviews and overall system designs. At first, this game seemed simple. It's not so simple.
CyberKnight Review In A Nutshell
Head Coach: "Now, for these reviews. Try to keep comments positive. Don't say 'You idiot! You clearly didn't think about this'. Instead try to keep it positive and say something like "Did you think about blah..?'.
Our design reviews take on different stages, and address different concerns. Different students and mentors have different experiences and will ask different questions pertinent to their area of knowledge and the current review stage. A basic design review addresses the mechanical system and robot requirements. Does it do what we want it to? Does it meet performance requirements? Is it robust enough to take hits? Does it resolve load and actuation forces? Next come system and operations questions. How does it package with other subsystems? How does it mount? How does the operator control it? What sensors does it use? (Does electrical/software know what the sensors are and how to use them?) We also review systems for parts and manufacturing. Can we physically build what has been designed? What does the Bill of Materials look like? Are parts capable of being produced in our shop efficiently with the equipment we have? Are necessary parts available for purchase? We look at systems from an operational perspective. (I look at systems from an operational perspective.) What steps and actuation's does this mechanism perform? How does this map to driver controls? Can the driver see feedback from this mechanism? What software extensions and limitations exist? Can we cancel an automated action? Each line of review questions follow lessons we have learned over the years, either ourselves, or from talking with friends on other teams.
For students, reviews can seem painful and negative. For mentors, reviews and feedback are likely a constant part of daily life and work. Students and mentors working together, trusting each others comments and feedback, will often produce the best results. The whole of the review process is to ensure that the final shipped product meets or exceeds the given standards. Each part is breathed on, edge cases are examined to the best of the designers and reviewers capabilities, and the final robot is well thought out and engineered.
Mechanics Versus Organics
Our top priority for the our first few years was always designing a complete robot, and then building it as fast as we can. We prioritized the process and the design. In 2016, the first time our drivers actually drove the complete robot was in Glacier Peak, during morning practice matches, mere hours before our first competition. Thank goodness we had excellent drivers with experience. In 2018 (and now 2019) we prioritized the schedule above all else. We will modify the design and the process to fit the schedule and get a complete robot into our drivers hands as early as possible.A good robot and great drive team will beat a great robot with a good drive team.
While this year may confine these next statements to the FRC boneyard, we are not yet at a point of fully autonomous robots in FRC. The best robot autonomous periods will outperform human drivers, and will be endlessly repeatable (given a static field with no interference from friends or foes). In 2018, how many drivers could score 4 cubes in the scale in 15 seconds? I know at least 5 robots that can. (The repeat-ability isn't 100% there, yet. It will be. Soon.) Over the course of a full match, with unknown elements thrown in (game pieces strewn all around, defense, etc) our drivers and operators <can> be much quicker at reading the data from the field in front of them. They <can> be much quicker at identifying the nearest game piece, acquiring, positioning, and scoring. The limit of our drive teams really exists only with their ability to play and practice. If you produce your robot one week early, your drivers can transform that time into on-field performance.
A great drive team with a good robot will beat a great robot with a good drive team. (Time to build a great robot and a great drive team.)
No comments:
Post a Comment