Scribd
Upload
28 views

Chapter 8 Dynamics and Control of Robots

This document discusses dynamics and control of robots. It covers inverse and forward dynamics, dynamic modeling using Lagrangian and Newton-Euler approaches, and control strategies including model-free independent joint control and model-based control. Model-based control uses the dynamic model of the robot to linearize the system and can include computed torque control, sliding mode control, and model predictive control.

Uploaded by

Siferaw Negash
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
28 views

Chapter 8 Dynamics and Control of Robots

This document discusses dynamics and control of robots. It covers inverse and forward dynamics, dynamic modeling using Lagrangian and Newton-Euler approaches, and control strategies including model-free independent joint control and model-based control. Model-based control uses the dynamic model of the robot to linearize the system and can include computed torque control, sliding mode control, and model predictive control.

Uploaded by

Siferaw Negash
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 30

Chapter 4

Dynamics and Control


Contents
• Inverse Dynamics,
• Forward Dynamics,
• Control
– Model free control
– Model based control
Dynamic model of a robot
• Forward dynamics
– Given forces/torques applied at the joints, find the
velocity and acceleration of the various links
– Analysis problem
• Inverse dynamics
– Given required acceleration and velocity of the
various links, find the required force/torque at the
joints
– Design problem
Dynamic model
• Two important approaches
– Euler-Lagrangian approach
– Newton-Euler approach
• The EL method gives closed form solution
which is useful for model based controller
design
• The NE method is iterative method which is
useful for computation
Dynamic model- LE method
• Lagrangian is defined as the difference between
total kinetic energy and total potential energy
• L=K-P
• Lagrangian form
d  L  L
    i
dt  q i  q i

• Where L is the Lagrangian,  is applied torque


or force at joint i, qi is the joint angle
Dynamic model-LE form
• The general form of robot dynamic model is

Mq q  Cq, q q  G q   


• Where
– M is n dimensional inertia matrix
– C is coriolis and centrifugal force/torque
– G is gravitational torque
–  is required force or torque
– q is joint angle
Derivation of dynamic model using Lagrange
method
• Consider an N DOF robot with links 0 to n and
joints 1 to n
• Let us assume the following notations
LE method
• The inertia tensor is given by

• We also need other parameters. Let us


assume the origin of link i-1 to be oi-1 and
origin of link i to be i as shown below
LE formulation
• Let us also assume we have this notations
LE form
• Steps to be followed
– Determine the frames from the Kinematic analysis
– From the shape of the links compute the inertia of each link
– Determine the kinetic energy of each link
• Kinetic energy is a quadratic function of joint velocities

– Determine the potential energies of each link


– Form the Lagrangian and compute the torques
Computation of kinetic energy
• The kinetic energy of a robotic link is the sum
of kinetic energy due to angular velocity and
translational velocity
• It will be computed in terms of the CG
• Kinetic energy of link is given by

• Since VG is velocity of center of mass, it can be


given by
LE form
• The inertia relation is given by

• Then the kinetic energy of link j becomes

• Then the potential energy will be calculated from

• Where L is the position vector from base frame


origin to origin of j frame
Dynamic model
• Properties of the dynamic model
– The mass inertia matrix is symmetrical
– The energy of link j is a function of (q1 to qj and
q1dot to qjdot)
– The inertia matrix Mji will be a function of qk+1 to
qn where k=min(I,j) and inertia parameters of link
r to n with r=max(I,j)
– The torque Ti is a function of inertia parameters
of link I to n
Dynamic model of 3DOF
• For the three DOF articulated robot, we can
design the dynamic model using the above
algorithm
• The position of the end effector wrt to the
joint angles is given by

• Let us first calculate the velocity kinematics


Dynamic model
• Differentiating the above equations

• Then rewriting in matrix form


Dynamic model of the 3DOF robot
• The matrix J is known as robot Jacobean
matrix.
• The determinant of the Jacobean matrix is an
important factor in the robot control

• The roots of the above equation are known as


singular points and the robot is not
controllable at this point
DM of 3DOF robot
• Following the above procedure we have
DM model
• The C matrix

• And G matrix will be


Dynamic model- NE form
• NE form is based on Newton’s second law
• The sum of all forces acting on a link of the
manipulator leads to a set of equations
• Two steps
– Forward recursion – calculate velocities and
accelerations
– Backward pass- compute the forces and moments
required
Control of robots
• In control of robots, we have to know
– The current position of the robot (link position and
velocities)
– The position of the target or the path to be
followed
• The controller to be implemented depends on
the performance required, the speed of
motion of the robot, available hardware and
so on
Manipulator control
• Individual joints of a manipulator are driven by
actuators that apply force or torque to cause
motion of the links
• The actuator commands that move the
manipulator and achieve the specified end
effector motion are provided by the control
system
Manipulator control
• The manipulator control system has two
inputs
– The set point values for the angles of the links
obtained from trajectory generators
– The actual value of the angle obtained from
sensors
Block diagram of manipulator control
Control strategies
• For slow motion – individual joints can be
moved one at a time and coupling is zero
• For fast motion- joints should move in
synchronism with each other – coupling cannot
be neglected
• Two methods
– Use independent controllers for each joint
– Use a model based controller that considers
coupling
Independent controllers for links
• Assumptions
– Link moves slowly
– Coupling is negligible or can be damped by
individual controllers
• Simple linear controllers are used for each leg
Independent controllers
• PID controllers are used for each leg
shafts

motor

k
lin
gear
Independent control
• The torque required at the motor shaft
  
 a  I a  a  Ba  a 

• Where
– First term is used to accelerate motor
– Second term is used overcome friction
– Third term is used to drive the link
–  is the gear ratio
Independent control
• Referring the link torque parameters
1 
 


a  I a a  Ba a  I  B 
 I   B 
a   I a  2 a   Ba  2 a
     
Model based control
• Computed torque control or feedback
linearization
– Uses the model of the robotic arm
– It linearizes the nonlinear dynamics
– Needs an ideal model
Model based control
• Feedback linearization controller
• Sliding mode controllers
• MPC

You might also like