Thanks to visit codestin.com
Credit goes to www.scribd.com

0% found this document useful (0 votes)
23 views9 pages

Control Report

The document details the control systems and dynamics of the ANT-X 2DoF Drone, focusing on thrust, torque generation, and linearization for control design. It describes the implementation of a PID controller for angular rate control and the cascade structure for position regulation, including the evaluation of performance through simulations and experiments. The results indicate effective control and tracking, though discrepancies due to real-world factors like sensor noise and actuator limitations are noted.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
23 views9 pages

Control Report

The document details the control systems and dynamics of the ANT-X 2DoF Drone, focusing on thrust, torque generation, and linearization for control design. It describes the implementation of a PID controller for angular rate control and the cascade structure for position regulation, including the evaluation of performance through simulations and experiments. The results indicate effective control and tracking, though discrepancies due to real-world factors like sensor noise and actuator limitations are noted.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 9

ELEC50004 – Control Systems

Control Systems Laboratory - 2DOF


Drone Report
Santos Santiago García Valls
02390906
March 2025
1. Introduction (Control_Lab1)
1.1

The total thrust produced by the quadrotor’s propellers is given by T c =−( T 1+T 2 +T 3+T 4 ), acting
downward along the body-frame z-axis. The negative sign indicates that thrust is directed opposite to the
positive z-axis of the drone’s frame.
Control torques around the roll (x-axis), pitch (y-axis), and yaw (z-axis) axes are generated by adjusting
individual propeller thrusts. The roll torque, Lc , is created by a difference in thrust between propellers
positioned along the x-axis. Specifically, propellers 1 and 4 contribute oppositely to those of 2 and 3,
b
leading to Lc = ∙(−T 1+T 2 +T 3 −T 4) , where b accounts for the arm length and aerodynamic effects. The
√2
1/ √ 2term shows the geometric arrangement of the propellers. By modifying thrust asymmetrically
between opposite propellers, a torque is applied along the x-axis, allowing roll control.

Similarly, pitch torque ( M c ) arises from thrust variations along the y-axis, involving propellers 1 and 3 in
b
opposition to 2 and 4. It is given by M c = ¿). This variation in thrust generates a net torque around the y-
√2
axis, enabling control over the pitch motion of the drone.

Yaw torque ( N c ) is produced by differences in the rotational speeds of the propellers, which result in a net
moment around the z-axis. This is described by N c =σ (T 1+T 2−T 3−T 4 ) where σ represents an
aerodynamic coefficient accounting for rotational drag and torque effects. Adjusting the speeds of opposite
propellers induces yaw motion, allowing the quadrotor to rotate about its vertical axis.

By controlling the speed of each propeller, the drone can regulate both T c and the torques ( Lc , M c , N c),
which together determine its movement and orientation.

1.2
Linearising the equations of motion means approximating a non-linear system with a linear one around a
specific operating point, making analysis and control design more manageable. In the case of the ANT-X
2DoF Drone, this is done by considering small deviations from a hovering state. Since quadrotors typically
operate near this condition, the complex non-linear attitude dynamics described by Euler’s equations can
be simplified. This allows the use of linear control methods, making attitude and stability control more
efficient.

1.3
The ANT-X 2DoF Drone updates control signals at 250 Hz and transmits telemetry data at 100 Hz. The
flight control unit (FCU) processes commands at the higher rate to maintain stability, while the lower rate
is used for communication with the ground control station.
In digital control, sampling effects must be considered when designing and tuning a controller. Since
control signals are held constant between updates due to the zero-order hold (ZOH) mechanism, a delay (
τ ZOH ) is introduced. However, this delay is usually insignificant compared to those caused by actuators,
sensors, and state estimation.
2. Identification (Control_Lab2)
2.1
Pitch motion in the ANT-X 2DoF Drone can be described using linearised equations that approximate its
behaviour near a stable operating point. The pitch angle ( θ ) changes over time according to its angular rate
(q ), expressed as θ̇=q . The angular acceleration is determined by the net torque acting on the system:
J θ q̇=M C + M E, where J θ is the moment of inertia, M C is the control torque generated by the propellers,
and M E is the exogenous torque due to external disturbances.

The exogenous torque M E is composed of three main contributions: gravitational torque M gE , which arises
due to the displacement of the drone's centre of mass relative to the pitch axis; frictional torque M fE , which
is caused by resistance at the shaft-bearings interface and has a damping effect; and aerodynamic torque
a
M E , which results from air resistance and also acts as a damping force. These external torques influence
the stability and behavior of the pitch motion.
To simplify control, the equations are linearised around a hovering state, producing θ̇=qand
2
q̇=−2 ξ ωn q−ωn θ+ μ M C , where ξ is the damping ratio, ω n is the natural frequency, and μ scales the
control torque.
Accurately identifying the model for pitch dynamics is crucial because it ensures that the designed
controller can effectively manage external disturbances and restore the system to its equilibrium state
swiftly and reliably. Without a precise model, the controller might fail to stabilise the drone or react
appropriately to changes in pitch motion.

2.2
To analyse the pitch dynamics of the drone a free response experiment was performed. This involved
briefly disturbing the system and recording how the pitch angular rate changed over time. The goal was to
extract key parameters that define the system’s behaviour, specifically the natural frequency ( ω n) and
damping ratio (ξ ).

By examining the response, the oscillation period ( T p) and settling time (T s) were measured, as they are
essential for characterizing the system’s dynamics. The recorded values were T p=1.7483 seconds and
T s=16 seconds.

From these, the natural frequency was determined as



ω n= =3.5939 rad/s, and the damping ratio was
Tp
4.6
calculated using ξ= =0.0800 . These values
T s ωn
indicate that the system is underdamped, meaning it
oscillates before settling into equilibrium.
2.3
The objective was to estimate the system gain ( μ) by
analyzing the peak values of both the simulated and
measured responses. From the experiment, the peak
measured response was 0.11rad /s, while the peak simulated response was 0.00129 rad / s. The gain ( μ)
0.11
was calculated as μ= =85.7085.
0.00129
As shown in Figure 1, there are noticeable discrepancies in amplitude and phase. The measured response
exhibits higher frequency noise and deviations, likely due to unmodeled dynamics or sensor noise.

3. Angular rate control (Control_Lab3)


3.1
The quadrotor's attitude control uses a cascade structure with an outer loop for attitude regulation and an
inner loop for angular rate control. The outer loop generates a reference angular velocity ( q d ) based on the
difference between the desired and actual pitch angles, while the inner loop ensures that the measured
angular velocity follows q d by adjusting the control torque ( M C). This torque influences the quadrotor’s
angular acceleration, governed by J θ q̇=M C, where J θ is the moment of inertia about the pitch axis. To
maintain stability, the inner loop must respond faster than the outer loop, requiring a higher bandwidth.
Typically, the outer loop operates at around 10 rad/s, while the inner loop is set between 15 and 30 rad/s,
ensuring that angular rate corrections occur before attitude adjustments, preventing interference between
the two control loops.
3.2
A PID controller is used for angular rate control in the quadrotor’s cascade structure, generating the control
torque M C based on the error between the desired and actual angular velocity. The standard PID control
K
law is given by R(s)=K p+ i + K d s, and K p , K i , and K d are the proportional, integral, and derivative gains,
s
respectively. Since the derivative term in this form can amplify high-frequency noise, it is modified with a
Ki Kd s
filtered derivative term: R(s)=K p+ + , where τ f is a small filtering constant. The proportional
s τ f s+1
gain ensures rapid corrections to deviations in angular velocity, the integral gain eliminates steady-state
error by compensating for accumulated discrepancies, and the derivative gain improves transient response
by reducing overshoot and oscillations. This
combination allows for accurate and stable
angular rate control while maintaining ease of
tuning and implementation.

3.3
The PID controller for angular rate control was
tuned through trial and error (Table 1), first in simulation
and then experimentally on the drone. The goal was to
ensure accurate tracking of angular rate setpoints while
minimizing overshoot and steady-state error.
In the simulations, run 1 provided a more stable response,
as seen in Figures 2 and 3. A lower derivative gain reduced
high-frequency oscillations, and the slightly higher
integral gain improved steady-state performance. Run 2,
with a slightly increased proportional gain, resulted in
faster response times but also introduced more oscillatory behaviour, likely due to the higher sensitivity to
rapid changes.
The experimental results in Figure 4 showed similar trends, with the PID controller successfully tracking
the setpoints, though with added noise and small deviations due to real-world factors such as sensor noise
and actuator imperfections. The general tracking behaviour was consistent with the simulations,
confirming that the selected gains provided effective control.
Comparing both runs, run 1 exhibited better stability and smoother tracking in both simulated and real-
world conditions, making it the more reliable choice. However, additional refinements, such as further
filtering or fine adjustments to the integral gain, could further improve performance and reduce residual
steady-state error.

3.4
The comparison between the simulated and
experimental pitch attitude responses shows a strong
correlation, demonstrating that the control model
effectively represents the system’s behaviour.
However, some discrepancies are noticeable,
particularly toward the end of the experiment, where
the measured attitude does not settle as smoothly as
in the simulation. This difference is likely due to
sensor noise, unmodeled aerodynamic disturbances,
or minor actuator imperfections affecting the real
system. Despite these deviations, the overall
agreement confirms that the PID controller achieves
reliable pitch attitude regulation, with the model
closely predicting the quadrotor’s dynamic
response.
3.5

Bandwidth Gm (dB) ϕ m (deg) ω gc (rad/s) ω pc (rad/s)


Approximated
Low 9.16 95.2 106 11.5
loop TF
Middle 6.95 57.8 89.4 32.6
High 5.3 41.5 81.6 41.8
Verification full
loop TF Low 9.16 98.9 107 12.7
(Continuous)
Middle 6.96 58.5 89.7 33
High 5.32 42 82 42.1
Verification full
loop TF Low 9.61 98.6 103 12.8
(Discrete)
Middle 6.9 57.7 87.7 33.5
High 5.25 41.3 81.2 42.6
Table 3: Attitude cascade control architecture

Loop shaping was used to design the angular rate control for crossover frequencies of 10 rad/s, 30 rad/s,
and 40 rad/s. The results include both the approximated loop transfer function and the verification of the
full-order system in continuous and discrete time.

Table 2 summarizes the controller parameters: proportional, integral, and derivative gains ( K p , K i , K d ),
designed and actual crossover frequencies (ω dc , ωc ), time constants (τ 1 , τ 2), control gain ( μ), and phase
margin (ϕ m ). As bandwidth increases, higher proportional and integral gains improve response speed but
reduce phase margin. The ω dc =30 rad/s configuration provides a good balance, with a phase margin of
57.8º, ensuring stability. Figure 6 shows
disturbance rejection across different frequencies.
At low frequencies, both continuous and discrete
responses remain near 0 dB, minimizing steady-
state error. The system attenuates disturbances
beyond a bandwidth of 70 rad/s. Minor differences
at high frequencies arise from discretization, but
the discrete controller closely follows the
continuous design.
While discretization introduces small deviations,
the phase and magnitude responses align well,
confirming accurate performance. The middle
bandwidth setting achieves the best trade-off
between response speed and stability. The low
bandwidth option has a higher phase margin but slower response, while the high bandwidth configuration
is faster but more sensitive to noise and instability.
3.6

For the 10 rad/s case, the response is slower, with some


steady-state error and phase lag. The measured response follows the simulated one but with added noise
due to disturbances. The pitch moment remains small, indicating minimal control effort. At 30 rad/s,
tracking improves, with the measured and simulated responses closely aligning. Phase lag is reduced, and
control effort increases slightly. This configuration provides a balance between response speed and
stability. With 40 rad/s, the response is fastest but more sensitive to noise. The simulated response follows
the setpoint well, while the measured response shows larger deviations, especially during transients. The
pitch moment is higher, increasing oscillations.
Measured responses consistently show more noise than simulations, particularly at higher bandwidths, due
to sensor limitations and real-world disturbances. The 30 rad/s setting provides the best trade-off,
maintaining accuracy with controlled effort and minimal instability.

4. Velocity and position control (Control_Lab5)


4.1

The cascade control structure for position regulation consists of an outer position control loop and an inner
velocity control loop. The position controller generates a velocity reference based on the position tracking
error, while the velocity controller converts this velocity error into a pitch reference that dictates the
quadrotor's motion.

The position error is defined as e x =x d −x , where x d is the desired position and x is the measured position.
The position controller, represented by Ropos, processes this error to produce a velocity reference
v d=R o ( s ) e x . The velocity error, given by e v =v d −v x, is then fed into the velocity controller Ri (s ),
pos pos

which outputs a pitch reference θd =Ripos ( s ) ev .This reference passes through the attitude dynamics T att (s ),
resulting in the actual pitch angle θ , which influences the horizontal velocity v x via Gv (s). The velocity is
then integrated through G x (s) to update position, closing the feedback loop.

To ensure stability, the frequency separation approach is used. The velocity loop operates at a higher
bandwidth than the position loop, allowing it to respond much faster. This makes the velocity loop appear
as an instantaneous response to the position controller, preventing interaction between the loops and
improving tracking performance. This separation ensures smooth position regulation while minimizing
overshoot and disturbances.

4.2

4.3

The performance of the outer-loop position controller was evaluated by comparing simulated and
experimental results for position, velocity, and pitch attitude. Figures 14, 15, and 16 illustrate the
differences between the expected and actual system behaviour.
The position response shows that both the simulation and measured data track the reference well, though
minor deviations occur during rapid changes, likely due
to unmodeled dynamics and external disturbances. The
velocity response exhibits greater oscillations and
overshoot in the experimental data compared to the
simulation, indicating the impact of sensor noise and
actuator limitations. The pitch attitude response
consistently shifts upward in the measured data,
suggesting a bias in pitch angle measurements. This
could result from sensor misalignment, calibration
errors, or slight variations in mass distribution. Factors
such as IMU placement within the quadrotor frame or
calibration inaccuracies could introduce these offsets.
Despite these discrepancies, the overall trends between
simulation and experiment remain consistent, validating
the controller's effectiveness while highlighting real-
world influences on system performance.

You might also like