Lecturer Notes: Robot Control Systems (Lecture 4)
Slide 1: Robot Control Systems (Title Slide with Diagram)
"Good afternoon. This slide presents an integrated functional diagram of a specialized
robotic control system. My purpose in showing this schematic is to emphasize that a control
system is not just a microcontroller; it's a hierarchical integration of components."
"The central block shows an AVR-controller Atmega328, representing the low level
processing unit for raw data from various sensors, such as an Inductive Proximity Sensor
and data from a specialized NDT (Non-Destructive Testing) Sensor."
"Crucially, note the presence of a dedicated PID controller block, which receives the actual
position signals from the Encoders attached to the motors. These encoders are the 'eyes' of
the control system, accurately reporting the position and velocity of every motor. The final
control commands are then sent to the motors, while auxiliary signals, like the Alarm and RF
module, communicate with the external environment. This schematic summarizes our goal
today: understanding how information flows to arrive at a control decision."
1 of 20
Slide 2: Introduction - The Brains Behind the Brawn
"We define a Control System as the 'brain' that directs a robot's mechanical body to
perform tasks accurately. It's the critical piece that converts geometric planning
(kinematics) and physical analysis (dynamics) into tangible mechanical action. As a student
in this program, you must master several aspects: first, the fundamental types of control
loops, like Open-loop vs. Closed-loop, and the crucial adaptive mechanism of the PID
unit; second, the mathematics, as we rely on Functions and Derivatives for path planning
and Numerical Methods to process the discrete data we acquire digitally; and third,
Communication factors, such as Bandwidth and Sampling Rate, which directly impact
performance, especially in systems involving force sensing or complex vision, where the
Robot Controller must process high-volume data in real-time."
2 of 20
Slide 3: Open-loop vs. Closed-loop Control: The Fundamental Divide
"This distinction represents the core of control systems: the choice between the simplicity
of no-feedback and the robustness of continuous feedback. It's fundamentally a decision
between blind, predefined execution (Open-loop) and adaptive, error-correcting
feedback (Closed-loop). In an Open-loop Control System, the signal flows one way, from
the controller to the actuator, without ever verifying if the output was correct—it’s a
predefined execution that ignores disturbances. Conversely, the Closed-loop Control
System features a feedback path from the motor back to the controller, allowing the
system to monitor and continuously correct for errors to maintain the desired
performance. The key differences are in adaptability, accuracy, and complexity; the
absence of feedback simplifies the system but critically compromises its accuracy and
ability to cope with external disturbances."
3 of 20
Slide 4: Open-loop Control: The Blind Executioner
"An Open-loop system is defined as command execution without output verification. The
classic analogy is baking without checking the oven's actual temperature or the dough's
readiness—the system assumes ideal conditions. Its primary benefits are simplicity, low
cost, and speed in predictable settings, making it common for fixed-sequence or timer-
based operations. However, its major drawbacks are a complete lack of adaptation to any
external disturbances, such as unexpected friction or a power drop, and its total reliance on
perfect initial calibration. The diagram clearly shows the linear flow: the Computer
Controller sends an order to the System Driver or Actuator, which executes it directly on the
Process or Robot without any sensing of the result."
4 of 20
Slide 5: Open-loop Diagram and Use
"Let's use the clothes drying system as a textbook example of open-loop control. The
Desired Time is the input, and the Timer sends electrical energy to the heating elements for
that set duration. Now, imagine a Disturbance, such as a much larger load of clothes than
usual, which affects the process; this disturbance is completely ignored. The system will
continue to operate for the preset time, potentially resulting in unsatisfactory Actual
Dryness without ever detecting the error. The diagram explicitly shows the unidirectional
control flow: Input $\rightarrow$ Controller $\rightarrow$ Process $\rightarrow$ Output,
with no feedback loop. This is the appropriate system only when the process is highly
predictable, low-cost, and involves fixed sequences."
5 of 20
Slide 6: Closed-loop Control: The Intelligent Adjuster
"In contrast to open-loop, a Closed-loop system is defined as a continuous feedback
system that minimizes error. The key feature is the Feedback loop which initiates the
cycle: Sensors $\rightarrow$ Error Calculation $\rightarrow$ Correction. This fundamentally
makes the system highly adaptable, accurate, and robust to changing conditions. The
trade-off is complexity; closed-loop systems require more intricate designs and can be
prone to instability if they aren't properly tuned. Common examples include robotic arm
position control and autonomous navigation systems. The diagram illustrates a conveyor
system where a Speed Sensor measures the belt's actual velocity and sends it back to the
Computer Controller, which compares it to the desired speed and adjusts the Actuator to
maintain a constant, stable velocity."
6 of 20
Slide 7: Closed-loop Diagram and Use
"This diagram represents the general architecture of any closed-loop control system. It
begins with the Setpoint or desired value, which is the target the output should achieve (e.g.,
$75.0^\circ$ C). The Controller compares the Setpoint with the value measured by the
Sensor via the feedback loop. Based on this Error—the difference between the desired and
actual value—the Controller sends an actuation command to modify the Process, driving
the error toward zero. This system is necessary when environments are unpredictable and
require high accuracy, as its primary strength is its ability to handle disturbances
effectively."
7 of 20
Slide 8: Open-loop vs. Closed-loop Comparison Table
"This comparative table summarizes the fundamental trade-offs every robotics engineer
must understand. Notice that Open-loop is less accurate and depends entirely on
calibration, while Closed-loop is more accurate because it continuously corrects errors.
Regarding stability, the Open-loop is inherently stable because there is no feedback to
cause oscillation. Conversely, the Closed-loop system can become unstable if it is not
tuned correctly, but its high adaptability and robustness in applications like mobile robot
navigation (e.g., Cruise Control) overwhelmingly justify the added cost and complexity."
8 of 20
Slide 9: PID Controller Fundamentals: The Backbone of Robot Control
"We now pivot to the most ubiquitous and powerful element of closed-loop control: the PID
Controller. PID stands for Proportional, Integral, and Derivative. This system is an error-
based adjustment algorithm that aims to minimize the Error between the setpoint and the
actual output by using three mathematical terms that operate in parallel. The error signal is
calculated at the summing point ($\Sigma$) and then sent to the three separate $P$, $I$,
and $D$ paths. Each path contributes a portion of the final control signal that is sent to the
process. PID units are widely used for controlling position, velocity, and temperature, and
are remarkably effective despite their relative mathematical simplicity."
9 of 20
Slide 10: PID Terms: Proportional (P)
"The P-term (Proportional) is defined as an immediate response that is directly
proportional to the current error. Mathematically, it is the Proportional constant ($K_p$)
multiplied by the current error ($e(t)$). It provides a quick reaction to deviations—think of
it as making an instantaneous steering correction while driving. Its effect is to reduce the
error quickly. However, a large value for $K_p$ can cause the system to overshoot the target,
leading to unwanted oscillation around the setpoint."
10 of 20
Slide 11: PID Terms: Integral (I)
"The I-term (Integral) is defined as the elimination of cumulative error. It accounts for the
history of the error, not just the error at the current moment. Mathematically, it is the
Integral constant ($K_i$) multiplied by the integral of the error over time. Its primary
function is to eliminate steady-state offsets—those small, persistent errors that the P-
term alone fails to correct entirely. Its effect is to grant the system high accuracy upon
settling, but it carries the risk of Integral Windup, a saturation condition that causes a
significant overreaction when the system recovers from saturation."
11 of 20
Slide 12: PID Terms: Derivative (D)
"The D-term (Derivative) is defined as predictive damping. It anticipates what will happen
based on the rate of change of the error. Mathematically, it is the Derivative constant
($K_d$) multiplied by the derivative of the error with respect to time. It provides damping
and reduces overshoot and oscillation, thereby improving the system's stability. It can be
analogized to anticipating deviations before they fully occur. However, the major drawback
of the D-term is that it is noise-sensitive; any high-frequency noise in the sensor signal is
heavily amplified by the differentiation process."
12 of 20
Slide 13: PID Equation and Tuning
"This is the unified PID Equation, where u(t) represents the combined control signal sent to
the actuator at time t:
$$\mathbf{u(t) = K_p e(t) + K_i \int e(t) dt + K_d \frac{de(t)}{dt}}$$
The process of Tuning is arguably the most critical step in applying PID control. It involves
adjusting the values of the coefficients Kp , Ki , and Kd to achieve the desired system
response. For instance, increasing Kp generally decreases the rise time but increases
overshoot. Ki improves steady-state accuracy, and Kd improves stability. Common tuning
methods include the Ziegler-Nichols method, alongside trial-and-error."
13 of 20
Slide 14: Feedback Signals: The Robot's Senses
"Feedback Signals are the real-time data that fuel the control loops. They essentially
represent the robot's senses. Sensors provide the essential state information, such as
position, velocity, and force. These can be Analog (continuous voltage/current) or Digital
(discrete binary data). The most crucial factors for feedback signal quality are low Noise and
low Latency. The cleaner and faster the signal, the more accurate the error calculation and
the better the system adaptation. The diagram shows how the Sensor measures the Actual
Output and sends it back to the Controller to generate the Error Signal. If this signal is
delayed or corrupted by noise, the controller operates on faulty or old information."
14 of 20
Slide 15: Functions and Graphs: Visualizing Motion
"In robotics, we use Functions and Graphs to describe and plan motion. Functions express
the mathematical relationships between variables, such as Position as a function of time,
$s(t)$. Graphs are used to plot the Motion Profile. The plot here shows a common profile
called an S-curve, where the motion consists of periods of constant acceleration, constant
velocity, and constant deceleration, ensuring smooth transitions. We use these functions
to plan the robot's trajectories, ensuring that the robot never attempts physically impossible
motions, such as instantaneous jumps in acceleration. This planning is necessary for
smooth, safe robotic arm operation."
15 of 20
Slide 16: Derivatives: Velocity and Acceleration
"Calculus is fundamental to understanding the rates of change in motion. We define
Velocity ($v$) as the first derivative of position with respect to time: $v = \frac{dp}{dt}$. And
Acceleration ($a$) as the second derivative of position (or the first derivative of velocity): $a
= \frac{d^2p}{dt^2}$. As we've seen, the derivative concept is crucial in the PID system,
where the D-term uses the rate of change of error $\frac{de(t)}{dt}$ to provide damping.
Understanding these derivatives is essential for path planning and dynamic model design.
The diagram also illustrates Jerk, the third derivative of position, which is controlled in high-
precision systems to ensure maximum smoothness of motion."
16 of 20
Slide 17: Numerical Approximation and Finite Differences
"Although we express motion using continuous derivatives, the computer operates on
discrete data samples. Therefore, we need Numerical Methods to bridge the gap between
continuous math and digital control. Finite Differences are discrete estimations of
derivatives. Instead of calculating the instantaneous $\frac{dp}{dt}$, we approximate it
using the change in position $(\Delta p)$ divided by the time interval $(\Delta t)$:
$\frac{\Delta p}{\Delta t} \approx \text{velocity}$. These approximations are relied upon
entirely in the digital control system. The accuracy of this approximation depends directly
on the Sampling Rate—the time interval between sensor readings. The smaller $\Delta t$ is,
the better the approximation to the true derivative."
17 of 20
Slide 18: Bandwidth and Sampling in Communication
"To ensure timely control, we must understand the role of Sampling and Bandwidth in
communication. Sampling is the rate at which data is captured from the sensors (measured
in Hertz - Hz). Bandwidth is the capacity of the communication link to transfer this data. The
Nyquist Criterion dictates that the sampling rate must be greater than or equal to twice
the highest frequency of the dynamic signal we are trying to measure. If the sampling rate
is too low, the phenomenon of Aliasing occurs, where the signal is distorted, and the robot
cannot react correctly to its environment, severely impacting its responsiveness and
stability."
18 of 20
Slide 19: Homework Assignment
"Your assignment for this week involves a practical application of the concepts we've
covered. It consists of Conceptual Questions that address real-world scenarios and
demonstrate the roles of the PID terms. Additionally, there are Numerical Problems
requiring you to estimate derivatives (velocity/acceleration) from discrete position data. The
Optional Simulation task for a PID control model is highly encouraged for those wishing to
gain hands-on experience in understanding how the values of the constants $K_p, K_i,
K_d$ affect system response."
19 of 20
Slide 20: Conclusion and Next Steps
"In summary, this week we've mastered the fundamentals of Control Systems, including
the distinction between open-loop and closed-loop, the mechanics of the powerful PID unit,
and the essential communication principles of sampling and bandwidth. We have
successfully linked kinematics and dynamics to real-world robotic operations. Our next
step in Week 5 is to explore Advanced Control concepts, such as adaptive and optimal
control. Please prepare for the next lecture on Robot Sensors and Data Extraction. Please
note the assignment due date is [October 25, 2025]."
20 of 20