outline

introduction
- background on RMLC, comparison to other control approaches
- background on unicycle project, comparison to other robot unicycles
- why we want to use RMLC on the unicycle

RMLC ("theory")
- description of RMLC (not too much detail necessary)
- details relevant to project, e.g. loss function, sample rate,
    extending state to 2nd order, quadratic controller

Hardware and Software
- summary of changes made
-- Arduino
-- gyro & accelerometer + I2C level converter
-- angle, position & speed sensing
-- flywheel stuff
-- voltage sensing
-- fall protection
-- error testing
-- cutoff switches and power switch

- sensing
-- description of MEMS gyros & accelerometers, noise levels and maths used
-- description of shaft encoders and speed sensing algorithm
-- description of self-centred position integration

- fall protection
-- skirt design
-- analysis of spin speed
-- reduced weight skirt failed, replaced with rope system

- reliability (or lack thereof)
-- HW self-test
-- I2C timeouts
-- fan belt problems, reduced max. motor command

Learning Results
- 2D unicycle w/ training wheels
- initial 3D run (balanced, but couldn't stabilise position)
- second 3D run (failed to balance)
-- possible reasons for second failure

Conclusion
Further Work
References
Evaluation of Risk Assessment
(Mathematical derivations?)