DEWBOT VIII IRI Post-Mortem
From DEW Robotics
The DEWBOT VIII IRI Post-Mortem was conducted on 25 July 2012. The team has begun to employ root cause analysis (Ishikawa) to enhance post-mortem improvement plans.
- Students: Andrew, Sasha, Doug, Molly, Kira, Mike, Pat D, Yahya, Ian, Jack, Ken
- Mentors: Clem McKown, Ben Kellom, Gary Deaver, Rita Wall, Siri Maley (in absentia: Julie Christopher)
- Facilitator: Siri Maley
- Autonomous - multi-revision error identified and corrected; consistent 2/2 afterwards
- Few Breakdowns - while some elements broke repeatedly (most notably the beaterbar and associated systems), we had no massive, diverse or fatal breakdowns. Nothing fully or even largely compromised our functions in any match.
- Drivetrain - notable among the systems that didn't break is the drivetrain. Refining pit processes has allowed the crew to keep on top of maintenance issues.
- Bridge Arm - while it wasn't used in any Triples (see "Improvements"), the arm's new robust mount ensured it worked to manipulate the bridge quickly and easily in every match--with no maintenance required!
- Shooter - shoot continued to hold up well and most missed shots were due to the feed into the shooter (i.e. the lift belt) rather than the shooter itself. New motor works well, though the PID control is not refined for the change, also affecting the shot percentage (though not as drastically).
- Teamwork - the pit crew/drive team rallied in the absence of the present-but-ill Pit Captain (Molly) for strong maintenance work and practice field improvement. Several students stepped up to fill pit roles
- 13th Seed! - 13 of 70 at the premier international all-star event of the season. We wish we higher (of course) and were close to seeding as high as 9th, but compared to just a few years or even one season ago, it's spectacular.
- Beaterbar (Mechanical) - the final beaterbar belt (roller to bottom pulley) slipped repeatedly, in at least 4 of 9 matches. Unfortunately, this severely incapacitated the robot's ball pickup and made us very inconsistent in both feeding and shooting roles. Belt length, trim and texture seem to have been resolved with a second replacing of the belt, but we will be monitoring and documenting this in practice. We're also examining lowing the final pulley so that the belt run is more parallel. (This was lowered some at competition, but requires further alteration.)
- Causal analysis: beaterbar stops/slows/stalls --> cord comes off pulley --> outside of both roller and belt are beat up & sticky --> belt may be stretching, second belt was not trimmed adequately --> belt is angled further downward than designed because beaterbar was lowered --> belt may be rubbing on chassis and/or bridge arm bracket
- Transverse Belt (Mechanical) - the transverse barrier drive/beater bar belt fell off twice (once apparently due to low tension, once from apparent over-correction), disabling both the beaterbar and the barrier drive. A tensioner is being installed on this belt.
- Lift Belt (Programming/Electrical) - the lift belt no longer auto-indexes balls, leading to inevitable jams. This appears to be a sensor placement issue and will be addressed.
- Causal analysis: repeated lift jams/belt losses --> must by operated manually as automated mode pulls first ball to top position immediately --> must run belt backwards to adjust to correct indexing positions --> combination drives belt off-center, pulls balls closer together --> jam and/or belt loss ensues. Automation failure --> false sensor readings (likely cause) --> flicker on third sensor (possible partial cause) --> examine sensor placement & power issues & sensor selection. Next year we will likely use different sensors entirely.
- Lift Belt (Mechanical) - Finally, the lift belt came all or most of the way off 3 times in addition to jamming twice more, rendering the lift inoperable for most of 2 matches and part of 2 more. All told we were incapacitated roughly 30-40% during the event, leading to issues of inconsistency with prospective alliance partners. Further improvement will be examined as necessary after auto-indexing is improved.
- Ball Path (Mechanical) - various factors seem to have moved the lift belt, shooter wheels, shooter cage, and lift rails out of alignment with each other. This was fixed partially at competition but will be further monitored in practice.
- Pit Personnel - having the required--and only the required--people in the pit has been a consistent problem for the team. At Duel on the Delaware we'll be implementing a new system in which the drive team asks (via whiteboard) for specific people directly after each match from the arena floor. (Design team members can also come down if they spot other problems).
- Pit Liaisons - an unprecedented emphasis on fixing autonomous and testing systems of the practice field along with the relative lack of repairs and the absence of Pit Captain Molly meant that the pit was often empty. At Duel on the Delaware, we will be implementing Pit "road signs" to direct prospective scouts to the practice field or indicate when our match is (if we're in queue). We're also examining pit display options that will be more scout-friendly. The closer personnel management will also ensure scouts always have someone knowledgeable present to talk to (or a sign indicating where/when to meet someone).
- Meeting Teams - while several students and mentors know other teams well and the drive team is well know for cooperation and smart play, our team as a whole would like closer collaboration with other teams. We're specifically interested in sitting and cheering with other teams at events (especially travel events), though there is a trade-off versus good scouting vantage points.
- Presenting the Robot - while most of the coaches and decision-makers on big teams understand and respect what we do (though our inconsistency at IRI prevented us from developing more non-local contacts), we have some gaps in our marketing.
- Hook Arm - We had 4 qual matches in which it made sense, as many partners had stingers and/or were long+overhang robots and required an outer position. In all 4, partners ranged from skeptical to very worried to insulted at our hook offer. The only team that agreed required that we were in the middle and "hooked" them on their belly pan, which is inline with their bumpers (meaning there's nothing to actually "hook" on). Unfortunately our last partner became incapacitated, and our hook partner came up backwards, leaving us to hook onto their intake (common given that the arm's position is within most team's intakes). Fortunately, it was very robust, as the mere potential of this occurrence vetoed 2 other potential hooks. Other declines came from related but more general fear of the hook, and a few pointed out that the hook was incompatible as it was within their dual-sided intake or their bumpers were mounted higher relative to their chassis. We're looking to mitigate this by displaying photographs of the hook arm in safely in use. We'd also like to practice with a long robot (namely DEWBOT VII Prime) in the middle.
- Scouting - we've embraced the need for a more serious and comprehensive scouting program. In quals, our current data is helpful but not definitive enough for each match. The data falls well short of alliance selection needs, a fate we have been very close to at both IRI and Worlds (it also would have been helpful at Lenape for the alliance that chose us). Scouting can bump us to the "next level": we were potentially one basket from being an alliance captain at both IRI and Worlds, and we lost one of our Lenape quarterfinals by only 4 points. We're currently preparing to implement a more comprehensive--and hopefully more fun--scouting program.
- Student Interest - most critically, some students have expressed lack of engagement at competitions and disappointment in their roles. Unfortunately we're getting to the point when we have more people than currently critical does. We're looking to:
- Implement a more interesting and effective data collection & strategy program
- Rotate students through the pit crew at the fall off-seasons
- Offer more exciting and engaging match watching experiences
- Further separate the pit crew and drive team to allow both to better focus at events and provide more opportunities
- Introduce volunteer opportunities for students (starting with an unrelated event, but potentially open to interested students at events we attend)
- Examine opportunities for other roles, including team liaisons