Non-linear manipulator control Flashcards
What are 3 ways to solve non-linear systems
-Local linearisation
-Moving linearisation
-Linearising control law
Why do we use the linearising control law to solve non-linear systems?
Linearising control law - Varying control gains for a critically damped system
-computing Kp such that the non-linear terms in the equation are cancelled out, leaving a linear model
Given a non linear force, how do we create a linear control law?
Replace the linear term of the open loop equation with the non-linear term
-find the model-based and servo portion equations under a unit mass
What are the practical considerations we need to make for robotic control systems?
-Time required to compute the model
-Sampling rate
-Lack of knowledge of parameters
What 3 control solutions take into consideration practical problems?
-Feedforward non-linear control
-Dual-rate torque implemenation
-Assuming a perfect model but with external noise
Explain feedforward non-linear control
-The dynamic model is outside the servo loop so model based torque is added at a slower rate
-the resulting error dynamic equation does not provide complete decoupling
-used as a starting point for robust control
Explain dual-rate torque implementation
-Expressing the dynamic model in terms of configuration space equation
-the functions can be computed via background process
when modelling external noise in a non-ideal situation, what is the system error dynamics?
error_2dot + kpe + kve_dot = 1/massvector x Tdist
Why do most industrial robots use simple control laws?
-Model-based control is computationally expensive
-good model parameters are hard to obtain
For industrial robot control systems, what controller do we use?
Individual joint PID control
-there is no model based portion, just servo based portion which contains PID
-therefore ß = 0
how can we determine control gains for the lowest resonant frequency?
Kv = wres
Kp = 1/4 (wres)^2
