Comparison of LTI and Event-Based Control for a Moving Cart with Quantized Position Measurements

In English Traditional linear time-invariant (LTI) control
design assumes that measurements are taken at regular time
intervals and have independent additive noise. A common
practical case that violates this assumption is the use of
encoders that give quantized position measurements; when the
quantization is appreciable the measurement noise is far from
LTI. This paper develops a simple event-based controller based
on simplifying a joint maximum a posteriori estimator, which is
applied to a moving cart with quantized position measurements.
The payoff for implementing the somewhat more complex
event-based controller is to drastically reduce the effect of
quantization noise in the experiments. A sequence of simpler
(LTI) to better adapted controllers are described and compared
according to experimental performance and implementation
complexity. Implementation issues on the microcontroller are
discussed.