Friday 2 September 2011

Grounding Research: Synthesis of Vex Design Principles

As previously promised, here is a synthesis of the Vex Inventors Guide - a basic, clear and concise introduction to the relevant engineering principles for robotics. The forum is a wealth of information also...

Apologies for the point form and brevity... but there is a lot of information in those pages, and to expand on everything would take way too much space...



>> structural
  •   rigidity
  •   centre of gravity (low and over "support polygon" pp. 2-33!)
    • weight of batteries + motors + chassis + tracks
    • weight of payload (ie carton of beer+ice)
  •  safety, bumpers etc to stop injury..
  •  stress / torsional forces / mounting points / material data sheets
    • bracing
    • suspension
  • exposure + vulnerability
Subsystem interactions:  
    
structural - motion 
  • needs to be codesigned, integrated for strength + support for integrals (motor, payload, batteries) 
  • handle rough, uneven terrain

structural - power
  • protect batteries, weight distribution  / centre of grave

structural - sensor
  • mounting and stabilising role
  • correct positioning for sensor tot work

structural - control
  • mounting and protection
  • antennae / rf shielding / range / reception

structural - logic
  • mount and protect (esp impact and h20)
  • wiring loom / connections robust and secure

               
>> motion:
  • speed vs torque
    • gearing / complexity / reliability / access to repair / maintain
    • drive gear / idler / compound gear / ratios 
  • materials for construction / durability
  • non gear, belt is better when motor and wheel is far apart
  • servo/pwm vs dc/h-bridge
    • build vs buy h-bridge or integrated chip
  • current draw (for batteries life and for motor controller integration
  • lasercut parts from cad? http://www.viz.tamu.edu/courses/tutorials/hunter/presentation1.htm
  • wheel size vs acceleration (smaller wheels = more torque)
    • stop / start of natural human gait, esp at party
    • safety, quick stop
    • more revs for regen braking etc
    • size for negotiation over obstacles (rocks, tussocks, hills)
    • clutch / stall protection (natural slip of belt?)
  • increase number of motors = cheap torque boost, but calibration..>?


motion - control - power subsystem interactions:

  • integrated microcontroller / h - bridge (sabretooth or robot claw)
    • regenerative braking / decelleration
    • regulated +5v dc for microprocessor power / protection
    • capacity / current handling capability of system matched



>> power
  • capacity + 20% (heavy contraption, never want to have to carry this beast)
  • discharge rates at full load / no load of motor / estimated Amp Hours (AH)
  • good metering / indication of low current / overcorrect  / regulation
    • possibilities of extending this with software control (similar to laptop power schemes, perhaps hack a laptop power monitor?)
  • li/po or deep cycle or motorcycle or scooter
    • shallow discharge = voltage drop
  • 5-12v or 24v (ohms law, high current/low voltage needs expensive copper) 
    • series / parallel benefits
  • ecology / green benefits of regenerative decelleration far outweighs the step outside the 'budget' construction
  • temperature 
    • batteries location within chassis 
    • heatsinking / fan for conroller


>> sensor
    eyes & ears, detect salient / important features of environment
  • reliability across all conditions expected
    • negative obstacles! cliffs, ditches
  • distance of operation / pickup polar patterns
  • analog / digital
    • analog is difficult to maintain constant / specific signal
    • digital is better for electrically 'noisy' conditions, but just hi/lo
    • combination / backup systems with different types of sensors to alleviate specific weaknesses
  • Tracking options...
    • IR proximity (pros n cons)
      • PWM for data coding (like IR remote control)
    • ultrasonic prox (see MaxBotix)
    • visibile light proximity http://letsmakerobots.com/node/1833
    • gps
      • accuracy? cheap options not hi res/fast enough updates...
    • rfid
      • signal strength (rough n ready!)
    • line detector concepts (hi/low or pattern detect ilke MS surface)
    • kinect depth mapping correlated to other sensor data (overkill, but interesting for future platforms? requires atx-mini-pc? robustness for doof?)
    • kinect mic array / positional audio capabilities (great for vox commands, could even incorporate similar tech to blob tracking code to reduce noise / predict direction)



 structure subsystem needs to accommodate all sensors
  • mount points
  • optimal operating range / dispersion / position
  • microcontroller / microprocessor handles regulated power supply (i.e BEC +5vDC supply from motor speed controllerr
  • perhaps hardwired bumper switch disconnects motors without microprocessor control for safety backup?
logic system requires cleanest data possible, with minimal processing time to make decisions quickly and accurately (necessary for following human)
audio recognition will require perhaps noise reduction / telstra technology ;)
control system will need backup (perhaps ability of manual operation in case of sensor subsystem failure) 
feedback (visual) from sensors and actuators integrated into aesthetics, but very useful for debug / troubleshooting, esp when 'back to nature' or out bush...




*phew*
-&c

No comments:

Post a Comment