Talk:DEWBOT VII

From DEW Robotics
Jump to: navigation, search

This is the start of DEWBOT VII.

First up is is our goals and objectives for 2010-2011 and the initial planning on how we are going to get there.

Goals

  1. Win 1 Regional Event
  2. Attend Championship event in St. Louis 27-30 April, 2011.

Objectives
To reach our goal we have a number of objectives that we need to achieve. Within those objectives will be a number of steps that we need to do. Objectives are grouped in major categories that match up with our sub-teams.

Drive Train

Dewbot VI

  1. Dewbot VI was the result of a summer and fall design project that resulted in our first Swerve drive robot. The pivot drive was an intense project and in the end successful. However the current design has much room for refinement. Now that we have gone down the swerve path, the next iteration should be easier.

Dewbot VII

  1. Dewbot VII work can begin by a redesign based on our lessons learned. Also, explore possible designs if we do not have the 4 steering motors next year. Work on encoders speed control for a total drive by wire system.

Gary Deaver 7/31/2010

  • Dewbot VI's pivots have stood the test of competition well. However, there were several problems that can be improved with some design changes. Any changes to the design should not compromise durability. The pivot drive was very expensive. With the budget constraints, cost reduction needs to be explored.

Starting at the bottom:

    • The wheel axles were made from grade 5 bolts we had in stock. The heads were cut off and the other end was threaded. Cutting the threads was time consuming and difficult. The threaded studs we bought had flats. There are studs available with out flats. MSC has them for 1.47 each part number 74363367. This will be a time saver.
    • The AndyMark plactraction wheels are fine. The durability of the conveyor belt treads could be better. Some research may come up with something better.The threaded brass inserts and screws tread attachment works well and should be used going forward. The Andymark AL sprockets and mounting system to the wheels works. All season there was a constant problem with the 2 shaft collars that hold the wheel in place. The shaft collars are the single set screw type and do not provide enough grip to the axle. We could eliminate the shaft collar on the sprocket side with an heavy duty e-clip or snap ring. The other side Could use a clamp on collar or a threaded clamp on collar. To keep costs down we could make these out of 1/2" polycarbonate.
    • Moving up to the 3/8" sprocket and miter gear drive shaft. Again, all through the season the set screws have loosened and require constant checking. The 9 tooth sprocket could be bonded on to the shaft with Loctite green gooo or epoxy, keeping the set screw and shaft key. Also the shaft can float if the set screws loosen. The addition of an e-clip or snap ring to the shaft next the bearing on the sprocket side would stop this. I've mentioned snap rings twice. Cutting groves in shafts on the lathe may need to be a trained skill. The cheap thrust bearings worked very well for both miter gears. Use them again. The plastic shaft spacer is a critical dimensioned piece. Can we make these in house?
      • The miter gears worked well as long as we have good alignment. The major big problem with these gears is their cost. We ordered finished bore gears. They are more than twice the cost of plain bore gears. We already have to cut key ways in the 9 tooth sprockets. For the miter gears the cost savings would justify having a marathon session to dill and tap set screws then broach key ways. As an example McMaster lists unfinished at $15.94 and finished bore at $37.59. Other suppliers list them as low as 11 something.

Mechanical Scoring

Programming

Programming the robot is the second most important thing that we do. You can't program a robot that does not work mechanically, but even a small amount of programming will improve the robot 300%.

The following are our programming objectives:

  1. A System Development Life Cycle - this is the methodology that we will use to create the software. There are many methodologies to choose from. We should choose the one that the programming mentor is the most comfortable with and has used. If there is no preferred choice then we should pick a lightweight one like Agile.
  2. System Development Tools - tools we need to develop the software
    1. Programming environment
    2. Software version control
      1. Presently looking at using Mercurial distributed version control and TortoiseHg Client
  3. Language Training
    1. RTOS - Real Time Operating System. What is a RTOS? What is different from RTOS programming compared to the classical structured programing model. How to instill a RTOS - Data Flow mindset to students from Vex to FRC. What do we need to do to develop an embedded mindset in our programmers. How do we avoid Toyota Syndrome and happy dance in our robot control systems?
      1. Labview is by design a RTOS data flow language. What is the data flow mindset? Looking at the code from early on to recently(end of April), the code has evolved into a more data flow style. How do we teach good LabVIEW style?
      2. C++ and VXWORX implements a RTOS environment. Is it to complex for us?
      3. Java can have RTOS like qualities if threading is implemented. Do we go down this path?


Once we have the robot rules and design. We can start working on the requirements and design.

Programming Tasks

  1. Select SLDC Methodology and do training
  2. Training on version control system
  3. RTOS Dataflow methodologies

Electrical

Drive / Tactical

Strategy/Scouting

Strategy/Scouting encompasses everything from detailed game analysis, to competition scouting software, to training match observers, strategists/coaches and drivers. It is perhaps the third most important task we undertake: a good robot with bad strategy will inevitably lose matches it should have won, while a bad robot with great strategy will often win matches it would otherwise lose. Our goal for this season is to improve the quality of our strategy, tactics and scouting information to the point of high importance, if not indispensability, in match play. (Note that "indispensability" is largely reserved for powerhouse teams who inevitably have very advanced scouting.) As such, our strategy/scouting plan for DEWBOT VII is as follows:

Develop & Implement a FIRST Strategic Mindset: This is the key to effective competition play. All the robot functions in the world and all the scouting data in the universe all comes to naught without strategy.
  1. Pre-Season: Analyze past games' rules and results to practice quickly and accurately identifying effective strategies, choke-holds, bottlenecks, and game-breaks. This experience is what makes game-breaker robots what they are. We're a relatively young team in this sense, which makes this research and practice invaluable.
  2. Pre-Season: Investigate other teams' approaches to game analysis. Some run miniFRC games in VEX, others have developed it into outstanding engineering design methodologies.
  3. Build-Season: Engage in rigorous game analysis starting the day after kickoff. This is an exceedingly important step that we, unfortunately, often under-develop. (It's also endorsed above all else by virtually every powerhouse team in FIRST.)
  4. Build-Season: Use this analysis as the basis for our functional design specifications, and employ them throughout the engineering design process.
  5. Build-Season: Develop scouting metric in tandem with game analysis. Not only, 'what's going to happen in this game?' and 'what are we going to do about it?', but 'how are we going to measure it?'
  6. Competition Season: Incorporate our game analysis and overall strategy into scouting, match-by-match tactics, and event strategy.
Enhance the Scouting System: The overall goals of scouting are, first, to determine match-by-qualification match strategy, and second (only possible after success in the first), assemble the best elimination alliance possible. As such, we have several different approaches open to us.
  • Pick the next highest team on the qualification rankings. This is by far the simplest, most passive, and unfortunately poorest choice on the list, as seeding score says little (nothing) about robot function and tactics compatibility. Many a good robot has lost in the quarterfinals due to this method.
  • Work jointly with another team's scouting system, contributing with our own scouts. This is in fact often the best-case scenario for young scouting teams. It provides insight into effective and experienced systems as well as the benefits of good scouting information and strategy.
  • Use another team's scouting system with our own scouts. This setup is rather rare. Few teams openly share their entire scouting system, though a few have been know to. More importantly, it deprives our own team, both scouts and strategists, of the opportunity to work and learn from experienced scouts and gain first-hand experience of an overall successful operation.
  • Create scouting software and use our team to scout. This is the ideal situation. However, it's only ideal when it works effectively, and this requires rather extensive experience.
As such, our plan for DEWBOT VII is to make friends with an advanced scouting team and work jointly with them. Of course, this depends heavily on what teams are at our events, so Plan B is to correspond long-distance with these teams for advice (throughout the year). We will also use this collaboration in addition to drive team experience to develop functional specifications and generate concept ideas for our own system. (Software development will advance as far as interest and ability carry us. This is likely a DEWBOT VIII goal.) Research and friend-making should be started in the pre-season and solidified based on build and competition season results.
Train the Teams: The best system in the world is useless without well-trained operators. Our plan for DEWBOT VII includes recruiting and training scouts, strategists, and drive team members. Scouts need to learn what and how valuable information should be observed, measured and recorded. Strategists need to practice developing this information into effective match plans and alliance selection lists. Coaches need to be able to work with drivers to observe and make on-the-spot decisions as well as also develop pre-match plans. Drivers need to be able to execute strategies and tactics as well as communicate effectively with each other and the coach. These jobs will likely overlap in personnel, but need to be staffed adequately enough to accomplish their jobs effectively. This is a pre-season, build-season, and competition season mission.

Media

While in a perfect world we would end up on top at the end of qualification rounds, in the event we don't, we need Plan B. We know that during alliance selection teams use their scouting. We also know from past experience that friends pick friends. Our objective for media team is very simply to have Sab-BOT-age known to every other team at the event.

We need a full court media push:

  • Sell the robot at the event
    • Teams that go from Pit to Pit talking about our robot
    • Buttons - portable advertising
    • Handouts - describe what our robot does
  • Sell the Sab-BOT-age brand
    • Local media - news articles
    • National media / websites
      • YouTube of our better efforts
    • Posts on Chief Delphi
      • Thoughtful, helpful posts
      • Downingtown and Sab-BOT-age listed in the signature

Program Management

The objectives listed above are difficult to achieve with our limited resources. We need to be very focused in our management efforts. The key is plan the work, work the plan.


See our other robots at FRC Team 1640