Introduction to Robotics | 2017 E.

Chapter 5

Robot Control

Robot control involves designing algorithms to achieve precise movement and interaction with
the environment. Different control strategies are used depending on the application and system
dynamics.

Figure; Block diagram for Mnipulator control system.

5.1 Natural Systems in Robot Control

Natural systems in robotics refer to control approaches inspired by biological organisms.

These systems emphasize adaptability, robustness, energy efficiency, and self-learning,
mimicking how humans and animals interact with their environment.

Natural systems often include:

• Biological motor control principles (e.g., muscle actuation)

• Sensory feedback mechanisms (e.g., vision, touch, proprioception)
• Neural-inspired learning (e.g., artificial intelligence in control)

Key Concepts in Natural Systems for Robotics

• Muscle-based control: Mimicking the contraction and relaxation of muscles in

biological organisms.
• Neural networks: Using networks of artificial neurons that resemble biological nervous
systems to make decisions.

KIOT 1
 Introduction to Robotics | 2017 E.C

• Biomechanical models: Understanding human or animal joint and limb movement and
applying similar principles to robotic actuators.

Natural systems typically involve nonlinear, dynamic control systems that rely heavily on
feedback mechanisms to adjust movements and maintain stability.

Example 1: Soft Robotics with Muscle-Like Actuation

Scenario:

• A soft robotic arm mimics human arm movement using pneumatic artificial muscles
(PAMs).
• The robot uses proprioception (self-sensing) to adjust grip force dynamically.

Application:

• Used in robot-assisted rehabilitation, where robotic exoskeletons help stroke patients

regain movement.

Example 2: Bipedal Robot Inspired by Human Walking

Scenario:

• A bipedal robot (e.g., Boston Dynamics’ Atlas) mimics human walking mechanics using
passive dynamics.
• The robot adjusts foot placement and balance based on sensory feedback.

Application:

• Used in humanoid robots for realistic movement and navigation on uneven terrain.

Example 3: Bio-Inspired Swarm Robotics

Scenario:

• Multiple small robots communicate and collaborate like ants or bees to achieve a task.
• Inspired by natural swarm intelligence, where each agent follows simple rules to achieve
complex behavior.

Application:

• Used in search-and-rescue missions where robots explore disaster zones collaboratively.

KIOT 2
 Introduction to Robotics | 2017 E.C

5.2 PID Control in Robot Control

PID (Proportional-Integral-Derivative) control is a widely used feedback control mechanism

in robotics. It adjusts motor inputs based on the difference between the desired and actual
positions (error) to achieve stable and accurate motion.

• Proportional (P): The control action is proportional to the current error.

• Integral (I): The control action is proportional to the accumulation (sum) of past errors.
• Derivative (D): The control action is proportional to the rate of change of the error.

PID control is used to control various systems in robotics, including robot arms, mobile robots,
and other actuators where precision is required.

The PID control law is given by:

Where:

• e(t) = Error (desired position – actual position)

• Kp = Proportional gain (reduces error quickly)
• Ki = Integral gain (eliminates steady-state error)
• Kd = Derivative gain (prevents overshooting by considering rate of change)

Key Features of PID Control

o Accurate Positioning – Ensures precise control of robotic joints.

o Stability – Prevents oscillations and overshooting.
o Error Correction – Compensates for disturbances and deviations.

Example: PID Control in a Robotic Arm (Position Control)

Scenario:

Consider a robotic arm with a single joint. The goal is to make the robot arm move to a desired
position using PID control. We want the end-effector to reach and maintain a desired angle (qd).

KIOT 3
 Introduction to Robotics | 2017 E.C

Step-by-Step Solution:
Step 1: Define the desired position

Let’s assume the desired position of the robot arm is qd = 90∘ (converted to radians: qd = π/2).

• The robot's current position is q(t) = 0 (starting at 0 radians).

• The error at time t is:

Step 2: Define the PID control law

We will use the PID controller to generate the control signal, which will be the desired torque at
the joint:

Assume the following parameters:

• Kp = 5
• Ki = 1
• Kd = 2

The robot's current position changes with the control signal τ(t), so the torque will adjust the joint
angle over time.

Example;

KIOT 4
 Introduction to Robotics | 2017 E.C

KIOT 5
 Introduction to Robotics | 2017 E.C

KIOT 6
 Introduction to Robotics | 2017 E.C

Application:

• Industrial robots for precise pick-and-place tasks.

• Robotic arms in medical surgery for accurate movement.

5.3 Joint-Space Dynamic Control in Robot Control

Joint-space dynamic control is a control approach used in robotics to manage the motion of a
robot by controlling the individual joints (or actuators) directly. It focuses on controlling the
robot in joint space, where the positions, velocities, and accelerations of each joint are the main
variables of interest. This type of control is essential for accurate positioning and smooth
movement of robot arms or other articulated systems.

In joint-space dynamic control, the robot’s motion is governed by its dynamics, including its
mass (inertia), damping, and stiffness. The control system typically involves feedback loops to
ensure the robot’s joints reach and maintain the desired positions, velocities, and accelerations
while accounting for forces like gravity and friction.

The Joint-Space Model

The dynamics of a robot arm (or any robotic system) can be expressed using Newton-Euler or
Lagrangian mechanics. In joint-space, the system dynamics can be described by:

Where:

KIOT 7
 Introduction to Robotics | 2017 E.C

In practice, these terms are used to model the robot's motion, and the control law must calculate
the required joint torques (τ) to achieve the desired behavior.

Control Objective
The objective is to design a controller that computes the joint torques τ to track a desired
trajectory qd(t), where qd(t) represents the desired joint angles as a function of time.

Example: Joint-Space Dynamic Control for a 2-DOF Robotic Arm

Given:

• A 2-DOF robotic arm with joint angles q1 and q2.

• The system dynamics are governed by the equation:

KIOT 8
 Introduction to Robotics | 2017 E.C

KIOT 9
 Introduction to Robotics | 2017 E.C

5.4 Independent-Joint Control in Robot Control

Independent-Joint Control (IJC) is a simplified control approach where each joint of a robot is
controlled separately, without considering interactions with other joints.

Unlike Joint-Space Dynamic Control, which accounts for forces like inertia and Coriolis
effects, IJC assumes that each joint operates independently, making control design easier.

Example; Independent-Joint Control for a 2-DOF Robot Arm

Given:

• A 2-DOF robotic arm with two joints.

• The control objective is to independently control the joints to track desired joint angles
over time.

Where:

KIOT 10
 Introduction to Robotics | 2017 E.C

KIOT 11
 Introduction to Robotics | 2017 E.C

Real-World Application:

• Basic industrial robots performing pick-and-place tasks.

• Lab automation robots that require simple joint movements.
• 3D printing robotic arms, where each axis is controlled separately.

Advantages of Independent-Joint Control

• Easy to design and implement

• Requires low computational power
• Stable when joint interactions are minimal

5.5 Task-Oriented Control in Robot Control

Task-Oriented Control focuses on controlling a robot’s end-effector or overall task rather than
its individual joints. Instead of controlling each joint independently, the system ensures that the
end-effector reaches a specific target in Cartesian space (x, y, z, orientation).

Mathematically, it works in Cartesian space rather than joint space, often using:

• Inverse Kinematics (IK): Computes joint angles for a given end-effector position.
• Operational Space Control: Controls forces and movements in the task space.

Example; Task-Oriented Control for a 2-DOF Robotic Arm

KIOT 12
 Introduction to Robotics | 2017 E.C

KIOT 13
 Introduction to Robotics | 2017 E.C

Real-World Application:

• Industrial pick-and-place robots (e.g., robotic arms in assembly lines).

• Surgical robots (e.g., da Vinci system for minimally invasive surgery).
• Humanoid robots performing object manipulation tasks.

KIOT 14
 Introduction to Robotics | 2017 E.C

Advantages of Task-Oriented Control

o More human-like control approach.

o Easier to program high-level robotic tasks.
o Better interaction with uncertain environments (e.g., force feedback for gripping).

5.6 Tracking Control in Robot

Tracking Control is a control technique where a robot follows a desired trajectory accurately
over time. The goal is to minimize tracking error by adjusting control inputs dynamically.

Tracking Control is widely used in:

o Robotic arms (precise movement for welding, assembly)
o Autonomous vehicles (self-driving cars, drones)
o Mobile robots (warehouse robots, delivery robots)
o Humanoid robots (walking and balancing)

Example; Tracking Control for a 2-DOF Robot Arm

Solution

To design a tracking control for the robot, we follow the steps outlined below:

KIOT 15
 Introduction to Robotics | 2017 E.C

KIOT 16
 Introduction to Robotics | 2017 E.C

6. Example Implementation and Result:

Suppose the desired trajectory in Cartesian space is:

KIOT 17
 Introduction to Robotics | 2017 E.C

5.7 Computed-Torque Control in Robot

Computed-Torque Control is a model-based control method used in robotic systems to achieve

precise motion tracking. It utilizes the robot’s dynamic model to compute the required control
torques that drive the robot’s joints to follow a desired trajectory.
This method compensates for:
o Inertia (Mass)
o Coriolis and Centrifugal forces
o Gravity effects

It is also called inverse dynamics control because it calculates the required torques based on the
robot's dynamics.

Consider a 2-DOF robotic arm with the following dynamics:

• The equations of motion for the robot are given by the following second-order differential
equation:

KIOT 18
 Introduction to Robotics | 2017 E.C

Example Problem Setup:

Let’s assume we have a simple 2-DOF robotic arm, where the dynamics are given as:

KIOT 19
 Introduction to Robotics | 2017 E.C

5.8 Learning Control in Robot Control

Learning Control (LC) is a technique that enables a robot to learn from experience in order to
improve its performance over time. Instead of relying solely on pre-programmed trajectories or
model-based control, Learning Control helps robots adapt to changes in the environment,
uncertainties, and disturbances.

Problem Setup:
Consider a robotic arm that is tasked with following a desired trajectory θd(t) (the desired joint
angles). The robot is assumed to have some level of noise or disturbance, and its dynamics may
not be completely known. The goal is to use Learning Control to improve the tracking
performance of the robotic arm over time.

The system is governed by the following equation of motion:

In Learning Control, we want to use the information from the past error between the desired
and actual trajectories to iteratively adjust the control parameters. The typical approach is to use
model-free learning, where the robot improves its performance based on feedback from the
environment rather than explicit modeling of the system dynamics.

Approach:

1. Error Definition: Define the tracking error at time t as:

KIOT 20
 Introduction to Robotics | 2017 E.C

Where θd(t)is the desired trajectory and θ(t) is the actual trajectory of the robot.

2. Learning Control Update: The Learning Control law adjusts the control inputs τ(t)
based on the past tracking errors. A simple model-free learning algorithm is to update the
control law as follows:
3. Learning Rate Update: The learning rate η(t) can be updated iteratively to improve
performance:

Where α is a constant learning factor. This equation adjusts the learning rate based on the history
of errors, allowing the robot to "learn" from its past mistakes.

Example: 2-DOF Robot Arm

Let’s assume we have a 2-DOF robotic arm with dynamics given by:

The desired trajectory is a sinusoidal trajectory θd(t), and the goal is to use Learning Control to
minimize the tracking error.

KIOT 21
 Introduction to Robotics | 2017 E.C

Types of Learning Control

• Iterative Learning Control (ILC): The robot repeats the task multiple times and
improves based on the error from previous iterations.
• Reinforcement Learning (RL): The robot learns through trial and error, receiving
rewards or penalties based on its actions.

5.9 Force Control in Robot Control

Force Control is a control strategy used in robotics to regulate the force exerted by the robot
on an object or the environment during interaction tasks. Unlike position or trajectory control,
which focuses on the robot's location, force control focuses on the force applied by the robot's
end-effector.

This technique is critical in tasks where contact forces are important, such as:

• Assembly tasks (insertion, fitting)

• Surgical robots (precise force application for cutting or tissue manipulation)
• Grinding, polishing, or welding

Force control can be achieved through various methods, including:

• Impedance Control: Controlling the robot’s dynamic response to forces.

• Admittance Control: Controlling the robot’s position based on external forces.
• Hybrid Control: Combining both position and force control.

Problem Setup:

Consider a robotic arm that needs to push an object in a straight line along a surface. The task is
to control the force applied at the end-effector of the robot to maintain a desired contact force
with the object, while also allowing the robot to move in the task space (e.g., pushing the object
while keeping the contact force constant).

Given:

• The robot's end-effector force F is related to the position error and velocity error through
a force control law.
• The robot's dynamics are described by:

KIOT 22
 Introduction to Robotics | 2017 E.C

Desired Task:

We want to control the force Fd applied by the end-effector while maintaining the robot's motion.
We will use impedance control (a common force control strategy) to achieve this. Impedance
control aims to regulate the robot's motion based on the interaction forces with the environment.

Impedance Control Overview:

Impedance control defines a dynamic relationship between the robot's motion and the forces it
applies to the environment. The basic idea is to define a desired impedance between the robot
and the environment (force, velocity, and position dynamics) and adjust the robot's behavior
based on force feedback.

The force control law can be written as:

Control Law for Force Control:

The goal of force control is to ensure that the force F applied at the robot’s end-effector matches
the desired force Fd. The robot’s motion should adapt to the force feedback, meaning that if the
robot applies too much force, it should slow down, and if it applies too little, it should speed up
or adjust its behavior.

KIOT 23
 Introduction to Robotics | 2017 E.C

The control law for force control can be written as:

KIOT 24