Meeting Notes 2011-09-23
Team members present: Curt, Stu, Aalyia, Dave, Oliver, Magy, Dr. Beth
- Presented system block diagram.
- Presented battery charging options. Desired abilities are to charge from AC or USB via the USB interface. Requires a chip MC1555 and a linear regulator. Requires safety precautions to interact with the Li-Ion battery, so a charge fall indicator and a current regulator are incorporated. Circuit gives 3.3V @ 200mA, 8 components at <$2. Li-Ion battery packs about 70x50x20mm costing consistently around $15-$30 per battery pack. Cell phone batteries might also be an option, but proprietary interfaces could be difficult to work with.
- SkyeTek SM-M9 with I²C interface probably won't work well for us given the previous team's experience. Better would be SPI or serial interfaces, since these are well specified and documented, require few pins, and should be easy to use. RS232 is a bit too pin-intensive.
- Flowchart of software routing and operation drafted. Still relatively abstract.
- Dealing with reduction of housing size. Dependent on input and feedback method. Factors include size of battery. Input keypad is relatively compact. Another possible input method would be a wheel with click feedback and set/enter button, similar in function to an iPod scroll wheel or a combination lock. Cell phone vibrators for feedback would be useful to save space. Selection of housing materials are also a concern in terms of weight.
- Heat dissipation: Consideration given to conduction, convection, and radiation. Conduction requires heat-conductive materials or heatsinks, while fans can be used for convection. Fans will probably be inconvenient because of power consumption and footprint.
- Researching servo motors. Still doing general research on servos. Controlling is typically done with microcontroller, but an analog support system using op-amps would put less stress on the MCU. Op-amps are small and cheap. Cell phone vibrators have been thrown around conceptually for a little while now, but might require a different control format. Feedback to provide error to the user and motion-based and direction-based feedback for giving directions should also be considered.
- Researching keypads and digital compass/gyroscopes. Will e-mail information.
- Most important consideration was the amount of documentation available. With preference to the Arduino interfaces available.
- Keypads run cheap.
- Gyroscopes run around $50 so that might be too expensive. Their main benefit is a bearing reading independent of three-dimensional movement, which might be unnecessary given the physical configuration. They do have problems like drift that would be awkward to work with. Digital compasses will likely be a better option. A three-axis accelerometer would also be cheap and easy to use (see P11011).
- Researching microcontroller units. Major considerations are size and geometry of the microcontroller, the ability to interface with the feedback methods (servos, etc.), availability of dev boards, and the amount of memory available on-chip (the maps will take at least 8 KB to store a full 1000 tags). In terms of these variables, the Arduino Nano 3.0 with 32KB flash/2KB SRAM/1 KB EEPROM is the first choice with emphasis on form factor, and the TI Stellaris 3L3748 with 128KB flash/64KB SRAM is a backup choice with emphasis on memory capacity.
- Dealing with ergonomics.
- Main consideration is comfort with the 8-hour wear time. Some intermediate material, like silicone gel or soft foam, should be put between the user's body and the main device.
- It might be good to move the input away from the output.
- Attachment options could include a spandex sleeve that conforms to the user's shape, or an adjustable strap like a belt or watch that can be set to size once.
- Human factors - the user needs to know that they're actually inputting information and need some feedback. Error prevention - there should be differentiation between buttons, and prompt feedback when a user wants to go to an invalid destination. Password-style redundant entry would work for input errors.
- Testing! Drop test, software test, learnability/training test, error output rate, equip/remove rate. Seeing people with a blindfold actually make good test subjects because they aren't used to working blind.
Week 4 Tasks
- Make slides
- Talk to EE professors about the project and find some subject-matter experts. Possibles: Saber and Sahin
- Investigate vibrational feedback and servos.
- Make a list of physical electronic components and dimensions for the ME team.
- Look up libraries for the SPI interface between the RFID antenna and the MCU, and write a rationale about selection of SPI over other interfaces.
- Bring Arduino.
- Look up digital compasses and accelerometers.
- Work on an implementation of Dijkstra's algorithm for the Arduino.
- Start working on compressing the maps to available memory on the MCU.
- Talk to CE professors about the project and find some subject-matter experts. Possibles:
- Talk to Dr. Marshall about ergonomics and input/output options.
- Input and output.
- Make a Gantt chart.
- More details on the feedback methods from last week's research.
- Heat calculations.
- Make decisions on feedback methods for the device and how they'll be used to direct the user. Outputs to communicate from the MCU and electronics include the compass heading, direction to travel, successful entry/error, etc.
All team meeting (Wednesday the 28th @ 8-9:30pm in the Senior Design floor)
- Create a Pugh chart.
- Polish up functional chart.