actuators

Article
Hybrid Control of Soft Robotic Manipulator
Arnau Garriga-Casanovas 1 , Fahim Shakib 2, *, Varell Ferrandy 1 and Enrico Franco 1, *

1 Mechatronics in Medicine Lab, Hamlyn Centre, Imperial College London, South Kensington,
London SW7 2AZ, UK; [email protected] (A.G.-C.); [email protected] (V.F.)
2 Department of Electrical and Electronic Engineering, Imperial College London, South Kensington,
London SW7 2AZ, UK
* Correspondence: [email protected] (F.S.); [email protected] (E.F.)

Abstract: Soft robotic manipulators consisting of serially stacked segments combine actuation and
structure in an integrated design. This design can be miniaturised while providing suitable actuation
for potential applications that may include endoluminal surgery and inspections in confined environ-
ments. The control of these robots, however, remains challenging, due to the difficulty in accurately
modelling the robots, in coping with their redundancies, and in solving their full inverse kinematics.
In this work, we explore a hybrid approach to control serial soft robotic manipulators that combines
machine learning (ML) to estimate the inverse kinematics with closed-loop control to compensate for
the remaining errors. For the ML part, we compare various approaches, including both kernel-based
learning and more general neural networks. We validate the selected ML model experimentally.
For the closed-loop control part, we first explore Jacobian formulations using both synthetic models
and numerical approximations from experimental data. We then implement integral control actions
using both these Jacobians, and evaluate them experimentally. In an experimental validation, we
demonstrate that the hybrid control approach achieves setpoint regulation in a robot with six inputs
and four outputs.

Keywords: soft robotics; machine learning; closed-loop control

1. Introduction
Citation: Garriga-Casanovas, A.; Soft robotic manipulators are attractive since they offer multiple degrees of freedom
Shakib, F.; Ferrandy, V.; Franco, E. (DOFs), a compliant structure, and the potential for miniaturisation [1,2]. This makes
Hybrid Control of Soft Robotic them potential candidates for applications that may include endoluminal surgery [3–5]
Manipulator. Actuators 2024, 13, 242. and industrial inspections [6–8]. An important part of these soft robots is actuated by
https://doi.org/10.3390/act13070242 pressurised fluids, typically air, and consists of a set of segments with 2–3 DOFs of actuation,
each stacked serially. One such example is [9,10], which we use as reference in this work to
Academic Editor: Han Yuan
explore control methods since it is a versatile soft robot that can be applied to a range of
Received: 27 May 2024 applications while offering higher force for a given diameter than similar designs.
Revised: 19 June 2024 The control of soft robots is a challenge due to their significant nonlinearity, continu-
Accepted: 25 June 2024 ous deformation along the arc length, and variability between prototypes. This last factor
Published: 27 June 2024 makes it difficult to create an accurate robot model and typical models contain significant
uncertainty. Accurate finite element models have been developed [11,12], but even small
variations in the manufacturing process due to practical tolerances lead to significant varia-
tions in robot behaviour. Despite these uncertainties, model-based closed-loop dynamic
Copyright: © 2024 by the authors.
control has been investigated by these authors [13]. However, it has so far only been
Licensee MDPI, Basel, Switzerland.
This article is an open access article
applied to control the robot in a predefined sub-region of the workspace, and while current
distributed under the terms and
models capture the robot behaviour locally, they fail to generalise to the full workspace.
conditions of the Creative Commons
More generally, models have been developed for soft robots [14], but they typically need
Attribution (CC BY) license (https:// to be tuned to the specific properties of each prototype, which vary due to manufacturing
creativecommons.org/licenses/by/ tolerances. Open-loop control has also been developed [15], but it leads to average errors
4.0/). of 10 mm and it cannot cope with disturbances. Stiffness control using impedance control

Actuators 2024, 13, 242. https://doi.org/10.3390/act13070242 https://www.mdpi.com/journal/actuators

Actuators 2024, 13, 242 2 of 20

has been developed for continuum robots [16], which can also include a gripper [17], but it
is applied to a one-segment robot that is tendon-driven rather than pneumatic.
Machine learning (ML) approaches for modelling or control have also been developed
for soft robot control [18–20] with approaches that can cope with hysteresis and nonstation-
ary behaviour [21], and approaches that perform model predictive control based on ML
for a fundamental model [22]. The main drawback of these approaches is that these tend
to require significant amounts of training data, typically in the order of ten thousand or
more datapoints. Reinforcement learning (RL) has also been explored for the control of soft
robots [23–25], but it also requires significant amounts of data. Soft robots, and in particular
the robot used in this work [9], have a limited life and noticeably degrade with use after
a few thousand cycles. Therefore, only a fraction of those cycles can be used to collect
experimental data for training. Hybrid approaches combining model predictive control
with iterative learning have also been proposed [26] although they are applied to a single-
segment robot. Similarly, nonparametric and online learning for a continuum robot [19]
and AI-based navigation of a bronchoscope [27] are applied to a single robot segment.
Potential applications such as surgery and confined space inspections typically require
precise robot control using the full workspace and multiple segments, and these operations
can last a few hours, which can be equivalent to a few thousand pressurisation cycles.
Thus, existing model-based control approaches [13] are generally not suitable given the full
workspace requirements. Pure ML approaches [18] and RL solutions are also not suitable
given the significant amounts of training data they require to achieve precise operation
in the full workspace, which deplete most of the life of the soft robot used in this work.
It should be noted that the specific requirements can vary, and in some applications that
require operating in a subregion of the workspace for hundreds of cycles, the existing
approaches can be appropriate. However, in general, a more robust and broadly applicable
control approach is desirable.
In this paper, we propose a hybrid approach to the setpoint control of the soft robot,
which combines machine learning and integral feedback control. Using machine learning,
we train an inverse kinematics model on experimentally obtained data to provide an
estimate of the required inputs to reach the desired setpoint. By integral feedback control,
we then regulate the pose of the soft robot resulting from the input estimated by the
inverse kinematics ML model to the setpoint. The work focuses on a robot actuated in
two segments, which has six pressure inputs and four pose coordinates as outputs. The
inverse kinematics is a mapping from the pose coordinates to the pressure inputs. We learn
this mapping using feedforward neural networks and kernel-based machine learning on
a limited-sized experimentally obtained dataset. Both neural network and kernel-based
models can effectively learn from limited-sized and sparse datasets, with minimal user
expertise and robust regularisation techniques to prevent overfitting [28,29].
We then validate these models experimentally. After that, we use these models to
generate an initial estimate of the pressure required to reach a desired pose. We then show
that with integral feedback control, we can reach the desired pose with satisfactory accuracy.
This work is essential for the soft robot to be used as a reliable actuator, and in the future
carry a surgical payload.
This work builds on our previous work [30], where a neural network is used for initial
pressure estimation followed by an integral control. However, our previous work [30] only
allows for robot operation in a small region of the workspace, where there is a high density
of training datapoints, and uses only four pressure inputs and three position outputs.
Moreover, our previous work relies on a model-free approach, where the relation between
pressure inputs and outputs is simply based on the sign of such relation, where an increase
in an input is directly mapped to an increase/decrease in an output based only on the sign.
This sign-based approach restricts its application to limited regions in the workspace.
The contributions of this work are the following: (i) a comprehensive exploration
and comparison of machine learning approaches (both feedforward neural networks and
kernel-based models) to learn the full direct and inverse kinematics of a full two-segment
Actuators 2024, 13, 242 3 of 20

robot with six pressure inputs and four pose outputs covering the full workspace and
using limited training data; (ii) an experimental validation of the selected machine learning
model, which can cope with redundancies in the kinematics; (iii) a hybrid control approach
combining machine learning with closed-loop control that can operate in the entire robot
workspace given that the control is based on the robot Jacobian; (iv) an exploration and
experimental comparison of both synthetic and data-based Jacobian approximations used
for control.
The hybrid approach is well suited to cope with the nonlinearities and redundancies
of the soft robot. The ML part learns the kinematics of each specific prototype using a
viable number of training datapoints, and serves as a first estimate of the required pressure
inputs accounting for the nonlinearities. The ML part also copes with redundancies, such
as six inputs and four pose outputs, and steers the inputs and outputs towards the desired
setpoint. The ML estimate does not lead to high precision since this would require an
excessively high number of datapoints, but the remaining errors are then compensated by
the integral feedback control. Moreover, the same training datapoints are used to estimate
the data-based Jacobian for feedback control, accounting for nonlinearities and behaviour
that can be unique to each prototype.
The paper is structured as follows. The control problem is formulated in Section 2. Ma-
chine learning approaches for both direct and inverse kinematics are proposed, validated,
and compared in Section 3. The hybrid closed-loop control is presented in Section 4 together
with experiments to validate the work. Finally, concluding remarks are summarised in
Section 5.

2. Problem Formulation
The aim of this work is to control a soft continuum robot consisting of two segments.
Each segment is designed following [31] with the robot design reported in [9]. In summary,
each segment has a tubular design with three internal chambers equally spaced at 120 degrees
and partition walls with a zig-zag configuration. The robot has an inextensible central tube
running along its centre, to which discs at the ends of the segments are glued. The chambers
in each segment are designed to expand when pressurised, pushing the central tube to one
side, and creating a morphing design that maximises the bending force [10].
Each segment bends approximately as a constant curvature arc. The two-segment
robot can thus be considered to bend in a piecewise constant curvature fashion when
no external loads are applied. Furthermore, each segment has three pressure inputs to
each of its chambers, and it offers two DOFs, which can be viewed as a bending direction
and a bending angle. Therefore, the two-segment robot has a total of four DOF outputs.
Consequently, there are two degrees of redundancy a priori between the six actuation
inputs and the four DOF outputs.
In practice, at least one chamber in each segment is always set to have zero pressure
since this is the chamber at the side towards which the robot bends. Applying pressure
in all three chambers of a segment simultaneously simply increases stiffness but does not
increase kinematic DOFs. In this work, we are not concerned with the robot stiffness, only
with the control of its position in free space.
The objective of this work is to perform setpoint regulation of the two-segment robot,
which has four DOFs as regulation outputs, with six pressure inputs and the condition that
at least one chamber in each segment should have approximately zero pressure. We set the
four DOFs as three position coordinates in Euclidean space X, Y,Z, and tip bending angle
θ. These position coordinates are collected in the vector x ⊤ := X, Y, Z, θ . The pressure
inputs are collected in a vector p⊤ := p1 , . . . , p6 .
 

The robot has a limited life of approximately 10,000 cycles. It also degrades with time,
leading to noticeable behaviour changes at roughly 5000 cycles (the specific value of cycles
changes significantly between prototypes, and this is an empirical approximation based on
four prototypes tested). The robot is also difficult to model due to manufacturing tolerances,
Actuators 2024, 13, 242 4 of 20

nonlinear material and deformation properties, internal deformation and buckling of its
internal partition walls, and hysteresis.
The approach proposed in this work is to use ML with a sparse dataset to learn the
robot kinematics in order to estimate the pressures required to reach the target, and combine
it with a closed-loop control based on an integral formulation on top. This approach suits
our robot since it requires a low number of training datapoints (fewer than 2000, which
ensures the robot has sufficient life for operation afterwards), and it is adaptive to the
particular behaviour of each robot prototype, including imperfections, nonlinearities, and
internal deformations, without requiring a detailed model of it. This work is focused on
the control of two segments to illustrate the approach, which is, however, generalizable to
three segments. The limiting factor is that training ML for a three-segment robot requires
additional datapoints, with the number of data growing exponentially with each segment.
Transfer learning will be explored in future work to reduce the amount of experimental
data required by relying on simulations.

3. Machine Learning for Direct and Inverse Kinematics

3.1. Pressure–Pose Dataset
We acquired a dataset that consists of N = 1369 datapoints of input pressures p ∈ R6
and realised poses x ∈ R4 :
D = { xk , pk }iN=1 . (1)
The dataset was acquired using structured pressure inputs at 0, 0.5, 1, 1.5, 1.8 bar,
applied to all six chambers using all relevant combinations of pressure inputs. The dataset
respected the condition that at least one chamber in each segment must always have zero
pressure. This dataset structure was selected given that its maximum size, i.e., N, is limited
by the robot life, and a well-structured data representation of all combinations of pressures
was needed to cover the full workspace.
The acquisition was conducted using the experimental setup shown in Figure 1. The
robot is 12 mm in its outer diameter, and it has a central catheter with a 2 mm working
channel [9]. The robot is made of three segments, each of them 26 mm long, and made
of Elastosil 4601M. Each segment has three equal chambers and a constant cross section
as reported in [9]. Each segment can bend up to 180 degrees and can apply lateral forces
up to 2.9 N when tested individually and held at the base [10]. However, in this work,
the bending is limited to approximately 90 degrees, and maximum force is approximately
1.5 N due to a limit on the maximum pressure to extend the robot lifespan to at least a few
thousand cycles. The supply tubes for distal segments pass through proximal segments.
The robot was hung upside down to eliminate gravity biases and it was pressurised
using proportional pressure regulators (Tecno Basic, Hoerbiger, Schongau, Germany) in
each chamber. These were controlled using a digital microcontroller (mbed NXP LPC1768,
Arm, Cambdridge, UK), which received the pressure commands via a serial interface from a
PC. A sensor was placed inside the central tube at the robot tip, and its pose was measured
using an electromagnetic tracker (Aurora, NDI, Waterloo, ON, Canada). The system
has an RMS of 0.7 mm, quoted by the manufacturer, NDI (Waterloo, ON, Canada). The
electromagnetic sensor was secured at the tip of the robot using blu tack, and a calibration
procedure was conducted at the beginning of each acquisition to ensure that the origin
of the coordinate frame is at the tip of the sensor when the robot is unpressurised and
the Y direction corresponds to the axis running along the centre of the robot at rest. A
Matlabscript (version R2017b) was used to set the robot pressure inputs and record the
robot pose for all the datapoints.
The resulting dataset is structured by three evenly spaced regions that reflect the
design with three chambers. The distribution of points in the three regions has similarities
but is not equal due to manufacturing tolerances and hysteresis in the robot.
Actuators 2024, 13, 242 5 of 20

Figure 1. Setup used for dataset acquisition and for testing. The depicted robot, used for experiments,
is fabricated with three segments, but only two segments were used in this work to investigate control,
which are the middle and distal segments.

3.2. Neural Network Architectures for Regression

We explored a range of neural networks (NN) to learn the direct kinematics (DK) and
inverse kinematics (IK) of the robot. We primarily focused on deep neural networks with a
fully connected architecture since we have limited a priori knowledge about the structure
of the DK and IK mappings.
The potential symmetries associated to the three equal chambers in each segment are
obscured by the manufacturing imperfections, internal deformation, and nonlinearities, and
thus the resulting DK and IK mappings cannot be assumed to have symmetries. This means
that convolutional neural network (CNN) architectures, which could exploit symmetries,
are not considered to provide additional benefits for this dataset. The limited number
of datapoints also means that residual network architectures are not appropriate since
deep networks with a high number of hidden layers in the tens would quickly overfit
the data. Residual networks improve the learning of identity mappings. However, given
the nonlinearity of the unknown function, we do not expect that residual networks will
provide a significant advantage over fully connected deep networks. Thus, we focused on
fully connected neural networks, using up to seven layers (five hidden layers in addition to
the input and output layers) and from 16 to 1024 neurons per layer.
We explored both small and large numbers of neurons relative to the dataset size to
reach the point of overfitting and exceed it. This was because of the sparse nature of our
dataset, which meant that some points in the dataset are likely to be unique in representing
features of the kinematics mappings. In the NNs, we used a rectified linear unit as an
activation function, and a Ridge regulariser with L2 = 0.01. In addition, we also added a
dropout of 10% in the training to prevent undesired overfitting.
All training was performed using a stochastic adaptive moment estimation (Adam)
optimiser 1000 epochs, and the mean squared error as the loss function. We also used a 10%
split between training and validation datasets in all the work, unless otherwise specified.
The data were shuffled at the beginning of every training, and 10% was reserved for testing.

3.3. Kernel-Based Learning for Function Estimation

In addition to neural networks, we also explored kernel-based learning for the learning
of the DK and IK mappings. Kernel-based learning has only a regularisation parameter
and typically only a few hyperparameters that need to be tuned, which enables an efficient
Actuators 2024, 13, 242 6 of 20

learning process. Furthermore, this regularisation parameter enables control over the
smoothness of the found mapping, which can enhance the generalisation capabilities of the
found model [28].
We first briefly recall the main principles behind kernel-based learning. Consider the
dataset {zk , wk }iN=1 , where w ∈ R is the set of observed outputs and z ∈ Z ⊆ Rnz the set of
(observed) inputs. Furthermore, consider a Hilbert space H characterised by a symmetric
positive semidefinite kernel function K. The space H defined by the kernel function K
and its hyperparameters has desirable properties such as completeness and a reproducing
property; see [28] for details. Kernel-based learning searches for the function g : Rnz → R
inside the space H that minimises the following cost functional:
" #
γ N 1
g = min ∑ (wk − ĝ(zk )) + 2 ∥ ĝ∥H .
ĝ∈H 2N i =1
2 2
(2)

In (2), the first term penalises mismatch in the data fit and the second term regularises
on the function complexity or smoothness. The regularisation constant γ ≥ 0 controls the
trade-off between these two objectives.
The minimiser g of (2) has a closed-form solution and is given by

N
g(·) = ∑ αi Kzi (·) (3)
i =1

α N ∈ R N being given by
 
with Kzi (·) := K (·, zi ) and α = α1 ...
  −1
1 1
α= K ZZ + γ−1 IN W, (4)
N N
⊤
where W = w1 . . . w N , IN ∈ R N × N is an identity matrix, and K ZZ ∈ R N × N is the


Gram matrix defined as K ZZ (i, j) := K (zi , z j ). If the output space w is multidimensional,

i.e., w ∈ Rnw , where nw > 1, we learn nw individual functions and concatenate them. The
function g in (3) is called the kernel-based model.
The quality of the data fit depends on the kernel, the kernel hyperparameters, and the
regularisation parameter γ. In the literature, a wide variety of kernels are proposed, among
others, linear, polynomial, rational, spline, and wavelet kernel functions; see [28]. Their
hyperparameters and the regularisation parameter γ can be tuned in various ways, e.g.,
by maximizing the so-called log marginal likelihood function; see [32] for details. In this
study, we focus on the squared-exponential (SE) kernel with automatic relevance detection,
which is defined as follows:
" #
nz 2
1 ( w ( i ) − w ( i ))
K (w1 , w2 ) := σ02 exp − ∑ 1 2
, (5)
2 i =1 σi2

where σ0 , . . . σnz are hyperparameters that need to be tuned. This kernel has the universal
approximation property [28]. The regularisation and hyperparameters are tuned using the
GPML toolbox [33], which maximises the log marginal likelihood function.
In the DK problem, the mapping g : R6 → R4 maps the pressure p ∈ R6 to the pose
x ∈ R4 in the dataset D in (1). Thus, in this problem, z = p, nz = 6, w = x, and nw = 4. In
the IK problem, the mapping g : R4 → R6 maps the pose x to the pressure p. Therefore,
z = x, nz = 4, w = p, and nw = 6.

3.4. Machine Learning for Direct Kinematics

We explored a range of fully connected NNs with a six-element input layer, a number
of hidden layers that ranged from one to three, and a four-element output layer. We used
2n neurons per layer, with a range of n = 4 to n = 9, and the regularisers described above.
Actuators 2024, 13, 242 7 of 20

Furthermore, we applied kernel-based learning as described in Section 3.3 using the SE

kernel in (5). The training of both the NN models as well as the kernel-based model was
performed on the same split in the dataset.
We found that the best NN model for the DK is composed of an input layer, three
hidden layers with 256, 128, 64 neurons, respectively, and an output layer, with an L2 = 0.01
regulariser applied to all weights in the network. The results of the evaluation of the DK
NN model on an unseen testing dataset are shown in Figure 2 and are represented by
the blue circles. In the same figure, we also depict the results of the evaluation of the
kernel-based DK model, represented by the red crosses, on the same unseen testing data.
The testing data are independent of the training data, but both of them are subsets of the
original dataset acquired, and are therefore obtained under similar operation conditions. If
the operation conditions of the robot change with respect to those of the training dataset,
for example, due to the addition of a constant load, we expect the performance of both the
NN model and kernel-based model to degrade in a similar manner since they both make
similar predictions on the kinematics and therefore they would generate similar errors.

Figure 2. Neural network (blue circles) and kernel-based (red crosses) models for direct kinematics
prediction on unseen testing dataset. The results show the actual pose versus the predicted pose. The
testing dataset is independent from the training dataset, but both datasets are acquired under similar
operation conditions as part of the original dataset.

To quantify the fitting, we define the coefficient of determination as

 
∑iN=1 | xi − ξ i |2
R2 := 1 − · 100%, (6)
 
2
∑iN=1 xi − 1
N ∑iN=1 xi

where xi denotes each actual value of the pose variables concatenated for the i-th datapoint,
N is the total number of points in the testing set concatenating the four variables, ξ i are the
predicted pose variables for the i-th datapoint, and | x |2 := x ⊤ x for x ∈ R4 . The R2 value
for the neural network model is R2 = 99.8% and for the kernel-based model is R2 = 99.9%.
Therefore, both machine learning methods successfully learned the DK mapping for all
four output pose variables.
Actuators 2024, 13, 242 8 of 20

3.5. Machine Learning for Inverse Kinematics

Learning the IK mapping is a more complex problem. The IK is a mapping from R4 to
R6 with the conditions described in Section 2. In particular, the machine learning models
for IK need to learn that at least one chamber in each segment must have approximately
zero pressure for each given pose. We hypothesised that the machine learning models
should be capable of learning this given that each region of the dataset is associated
with zero pressures in specific inputs, which is the result of the geometric bending of
the robot to reach each region. Thus, the IK machine learning models would map each
region in R4 to a hyperplane in R6 that corresponds to zeros in specific pressure inputs, if
trained appropriately.
A potential issue arises at the boundaries between regions that have zero pressure in
different inputs. However, knowledge of the behaviour of the robot, described in Section 2,
indicates that boundaries are not expected to be a problem. For a given segment, the
transition from bending in a direction that corresponds to zero pressure in one chamber
to another direction that corresponds to zero pressure in the adjacent chamber is to pass
through a configuration where both chambers have zero pressure simultaneously. Thus, a
continuous transition between regions is expected, which should be reflected in the data
and thus can be learned.
Given these considerations, we explored a range of fully connected NN models to
learn the IK. All NN models explored consisted of an input layer with four elements,
a set of hidden layers that ranged from two to five, and an output layer with the six
predicted pressures. We used 2n neurons per layer with a range of n = 4 to n = 10, and the
regularisers described above. Furthermore, we also learned a kernel-based IK model using
the approach in Section 3.3.
The best IK NN model was found to consist of an input layer, three hidden layers with
64 neurons each, and an output layer, with a L2 = 0.01 regulariser. The corresponding pre-
diction of this model on unseen testing data is shown in Figure 3 (blue circles). This figure
also depicts the predictions of the kernel-based IK model (red crosses) on the same testing
dataset. Analogous to (6), the coefficient of determination R2 quantifies the performance of
the IK models. When evaluating the unseen testing dataset, the IK NN model achieved
R2 = 79.8%, whereas the kernel-based IK model achieved R2 = 78.0%.
These results confirm that the explored machine learning models can learn the IK
mapping and generally predict which pressure inputs should have approximately zero
pressure. However, these models do not exactly predict zero pressure in all occasions,
and in some cases, these models predict a negative pressure instead of zero. This is due
to the fact that the data-driven models estimate the inverse kinematics function from a
limited-sized and sparse dataset in an unconstrained fashion. In the regions where negative
pressures are predicted, the inverse kinematics mapping is not accurate and provides, on
some occasions, negative pressure values. By saturating negative pressures to zero and
considering that the robot bending is nearly negligible at pressures below 0.5 bar, the IK
NN model correctly predicts nearly zero pressure (below 0.5 bar) in 92.6% of all cases in
the test dataset, whereas this score is 91.9% for the kernel-based IK model.
Actuators 2024, 13, 242 9 of 20

Figure 3. Neural network (blue circles) and kernel-based (red crosses) models for inverse kinematics
prediction on unseen testing dataset. The results show the actual pressures versus the predicted pressures.

3.6. Inverse-Direct Machine Learning Models

Machine learning for the IK mappings shows a significantly lower coefficient of
determination than machine learning for the DK mapping. We developed a k-nearest
neighbours (k-NN) algorithm to investigate the potential source. For a given pose, the
algorithm finds the three nearest points in the training dataset and the corresponding
pressures. This reveals that in over 25% of instances, for a given pose, there are two or more
points in the training data within 1 mm Euclidean distance. However, the pressures used to
reach those points present significant differences, both in terms of values, which can vary
by over 1 bar in the chambers, and in terms of the inputs that are set to zero pressure. This
indicates that pressure inputs that are significantly different can lead to very similar given
poses. Consequently, despite the spread in the predicted IK pressures, the values provided
by the machine learning models are expected to lead the robot near the desired pose.
To evaluate the spread in the predictions of the learned IK models, we combine the
IK models together with the DK model to create an IK-DK model. Thus, for a given pose,
we run the IK model to find the pressures required to reach the desired pose, and we then
input the pressures to the DK model to evaluate the resulting pose. If the IK models and
the DK models are accurate, then the IK-DK model should be an identity mapping.
The IK and DK models are trained independently. Therefore, the IK-DK will return a
point close to the original pose if both the IK and DK models are accurate representations
of the true IK and DK mappings, but it will also highlight errors in each model if these
individual mappings are not learned accurately.
The results of the IK-DK models are shown in Figure 4 for the NN models (blue circles)
and the kernel-based models (red crosses). For the NN models, the combined coefficient
of determination is R2 = 98.1%, whereas this score is 97.3% for the kernel-based IK-DK
model. These results indicates that the IK NN model and kernel-based IK model are a good
approximation of the true IK mapping. Furthermore, we can conclude that the deviations
shown in Figure 3 in terms of pressures, in fact, do not affect the resulting pose significantly
Actuators 2024, 13, 242 10 of 20

since they converge to a resulting pose that is close to the desired one. It should be noted,
however, that the coefficients of determination of the IK-DK are slightly lower than those
of the DK in Section 3.4 (R2 = 99.9% and R2 = 99.8%, respectively). This is due to the fact
that, even though poses of points nearby converge, a slight error is introduced by the IK
since the significant differences in pressures to reach two nearby points mean that the IK
mapping has to fit abrupt changes that create a degree of error. This results in a slight drop
in the coefficients of determination for the IK-DK.

Figure 4. Neural network (blue circles) and kernel-based (red crosses) models for the composed
mapping of direct kinematics prediction evaluated on the inverse kinematics predictions on unseen
testing dataset. The results show the desired pose versus the predicted pose.

3.7. Experimental Validation of the IK Neural Network Model

The DK and IK kernel-based models show similar performance to the NN models.
Therefore, only the IK NN model selected in Section 3.5 was validated experimentally. The
testing dataset poses were input into the IK NN model, which predicted the pressures to
reach them. The pressures were then applied to the robot using the same setup used for
data acquisition. The robot pose was measured and recorded using the EM sensor.
The results of the measured poses reached by the robot versus the desired poses are
shown in Figure 5. The results validate the IK NN model, but they also show that there is a
degree of error in the experimental pose reached that is greater than in the IK-DK evaluated
numerically in Section 3.6. The experimental coefficient of determination is 94.87%. We
attribute this to the fact that the robot exhibits a memory effect, which is the combination
of hysteresis, and the fact that deformation in the chambers does not flow easily from one
pose to the next due to internal friction and buckling. The training data and the machine
learning models capture this memory effect. When performing IK-DK numerically on the
testing dataset, which is a subset of the full dataset gathered at once, both the DK model
and the IK model contain a similar effect of memory in their mappings that they learned
from the same training dataset. However, when experimentally testing the IK NN model,
the robot is commanded to reach points in a different order than in the training dataset, and
that means that the memory effect creates deviations that are different from those captured
by the NN models. Thus, the deviations due to memory effect increase the errors when
comparing the experimental validation of the IK NN model with R2 = 94.87% and the
Actuators 2024, 13, 242 11 of 20

numerical IK-DK machine learning models with R2 = 98.1% and 97.3% respectively. It
should be noted that the scaling in the y-axis plot in Figure 5 is different from the other axes
since the total range of motion is lower in this axis. This makes the errors more apparent in
the y-axis, but the absolute errors are similar to those in the other axes.

Figure 5. Experimental validation of the IK NN model when tested on the robot.

4. Hybrid Closed-Loop Control

The results in the previous section show that the IK NN is capable of approximating the
robot to a pose within a few millimetres of the desired setpoint. A closed-loop formulation
based on an integral is proposed in this section to compensate for the remaining error and
reach the desired setpoint.
A key question is how to map the error in the four pose outputs to the six pressure
inputs. In the following subsections, we explore two approaches based on either an
experimental or synthetic Jacobian.
In both approaches, a hybrid closed-loop control is proposed based on the IK NN
model and an integral controller that is formulated for setpoint regulation of the robot
using the Jacobian pseudo-inverse. The proposed control is
Z
U = IK ( xt ) + k i J−1 ( xc ) edt + J−1 ( xc )k p e (7)

where U is the control input consisting of six pressures (p1 , p2 , p3 , p4 , p5 , p6 ), e is the tip pose
error (in x, y, z, θ), k i is the integral parameter, k p is the proportional term parameter, and
J−1 ( xc ) denotes the Jacobian pseudo-inverse using the (4 × 6) matrix terms corresponding
to the four tip pose variables x, y, z, θ differentiated with respect to all six pressures, and
evaluated at the current robot pose xc . The prediction by the IK NN model is denoted by
IK ( xt ) and provides approximate pressures to reach the desired pose. IK ( xt ) plays its main
role at the beginning of setpoint regulation as a first estimate of the pressures, after which
the integral term becomes increasingly prominent to compensate for the remaining error.
The control is suitable as long as there is no saturation, and the dataset is sufficiently
dense. We use unwinding of the integral to mitigate the effect of saturation [34].
The various approaches to estimate J−1 are described in the following subsections.
Actuators 2024, 13, 242 12 of 20

4.1. Experimental Jacobian Control

The first method explored to estimate the Jacobian is based on an experimental ap-
proximation of the Jacobian. This is extracted from the full dataset gathered experimentally,
and thus is intended to capture the behaviour of the specific prototype.
To estimate the Jacobian at a given point, first a k-Nearest Neighbours (kNN) algorithm
is used to find the nearest datapoint in the full dataset. Then, the partial derivatives of
the six pressures with respect to pose, i.e., ∆pi /∆Xi , are estimated for that point found.
To estimate the partial derivatives, we search again the full dataset to find the datapoints
corresponding to an increment in each of the pressures p1 to p6 . Since the dataset is
structured, there is always a datapoint corresponding to either an increment of 0.5 bar or
0.8 bar. For datapoints at the edge of the dataset, which are obtained at a 1.8 bar pressure,
the partial derivative is estimated based on the datapoints corresponding to a decrement
of pressure. The partial derivatives are finally assembled in matrix form and normalised,
yielding the desired estimate of the Jacobian inverse J− 1
e .
In order to use Je in (7), it needs to be evaluated near xc . In this approach, J−
− 1 1
e is
evaluated at the target xt to ensure that it remains stable. This assumes that J− 1
e ( xt ) is
representative of the current Jacobian, and is valid as long as the IK NN brings the robot
sufficiently close to the target. Thus, in this approach, the controller (7) is implemented
using J−1 ( xc ) = J− 1
e ( x t ).
The controller (7) with J−1 ( xc ) = J− 1
e ( xt ) was experimentally tested using the setup
described in Section 3.1 but using a Python script in an environment with Tensorflow
to implement the IK NN and the closed-loop control. The setpoint was selected to be
x = −4 mm, y = −0.5 mm, z = 1 mm, θ = 6◦ , which is a representative point inside the
workspace and not close to saturation. A value of k i = 0.2 was used, and the proportional
term was not used, i.e., k p = 0. The value of k i was selected by experimentally exploring a
range of values (from k i = 2 based on previous works [13] to k i = 0.1) and choosing the
one that led to better performance. Higher k i leads to faster response and adaptation to
disturbances but it can also result in overshoot and larger oscillations. There is also a risk
that the robot reaches saturation during overshoot and oscillations, making higher k i less
desirable. Lower k i offer slower response times, but they reduce overshoot and oscillations.
In this work, a compromise value of k i = 0.2 was selected since it reaches the target in
approximately 2 s while avoiding oscillations or overshoot.
The experimental results with controller (7) are shown in Figure 6. As can be seen,
the robot tends to the desired pose. There are low oscillations caused by the large value of
k i . This value can be reduced to mitigate oscillations, but this results in a slow response,
creating a trade-off between oscillatory behaviour and settling time. Another potential
solution to reduce oscillations is to add a dissipative term to the control law (7), such as
a derivative term proportional to the velocity of the robot. However, velocity estimation
with our experimental setup can have limited precision and lead to local errors. Future
work will explore methods such as filtering to reliably add dissipative terms to mitigate
oscillations. Overall, the experimental approximation of the Jacobian yields a controller
that converges as long as the IK NN provides a good first approximation of the pressures
required to reach the target.
A potential drawback of the experimental Jacobian is that it can be inaccurate at poses
far away from the target pose xt (typically a few millimetres). As can be seen in Figure 5,
the error of the IK NN can be significant for some points, in which case controller (7) does
not converge. This is particularly an issue when the IK NN model initially sets pressures
that move the robot to a pose distant from the target. Furthermore, the Jacobian can be
inaccurate in regions of the dataset where the Jacobian is distorted due to the robot’s
memory effect.
Actuators 2024, 13, 242 13 of 20

Figure 6. Experimental results of setpoint regulation using a hybrid control combining IK NN and an
integral action based on an experimental estimate of the Jacobian.

4.2. Synthetic Jacobian

A synthetic Jacobian is explored in this section to avoid the issues of distorted and
abruptly changing Jacobian caused by the experimental estimation in the previous section.
The synthetic Jacobian is derived by assuming piecewise constant curvature (PCC) bending
of the robot segments. We use a model relating the bending angle and bending plane of
each segment to the pressure applied to its chambers, and combine it with robot kinematics
of the two segments stacked.
The segment model used is based on [13], and is
√
3( p − p )
tan(ϕ) = 2p2 − p1 3 − p3
1
q (8)
θ = 1k p21 + p22 + p23 − p1 p3 − p2 p3 − p1 p2
Actuators 2024, 13, 242 14 of 20

where θ is the segment bending angle, ϕ is the bending plane of the segment, p1 , p2 , p3
are the pressures applied to the three chambers of the segment, and k is a parameter
representing the segment structural stiffness. The same model is used for both robot
segments, with k = 5 bar/rad.
The robot kinematics are based on [35]. They are used to find the direct kinematics
relating the tip pose to the arc parameters of the robot segments as

x, y, z, α, θ, γ = f (ϕ1 , θ1 , ϕ2 , θ2 , l1 , l2 ) (9)

Here x, y, z represent the robot tip position, α, θ, γ are the ZYZ Euler angles of the robot
tip orientation, all relative to a reference frame positioned at the robot base with the Z vector
co-axial with the robot centreline when not pressurised, ϕi , θi are the bending plan and
angle of segment i, and l1 and l2 are parameters representing the lengths of the segments,
respectively. In this work, l1 = l2 = 25 mm. The specific function f is not reproduced here
for length reasons, but it can be obtained from [35] using the homogeneous transformation
of each segment

s2ϕi 1 − cθi + cθi

sϕi cϕi cθi − 1 cϕi sθi σi sθi /2 cϕi
 
c2ϕi 1 − cθi + cθi



 sϕ cϕ cθ − 1 sϕi sθi σi sθi /2 sϕi 
i i
Ti = 
 i  (10)
−cϕi sθi −sϕi sθi c θi σi cθi /2 
0 0 0 1

2l sin(θ /2)
where σi = i θ i , and sw and cw denote the sine and cosine of w, respectively.
i
The total transformation for the robot is

T t = T1 T2 (11)

Then, the tip x, y, z can be directly obtained from (11), and the ZYZ Euler tip angles
can be obtained, e.g., see [36], as
Tt23 Tt13
α = atan2(q
sin θ , sin θ )
θ = atan2( Tt231 + Tt232 , Tt33 ) (12)
Tt32 Tt31
γ= atan2( sin θ , − sin θ )

Finally, substituting the segment model (8) into the kinematics (9) yields

x, y, z, α, θ, γ = f ( p1 , p2 , p3 , p4 , p5 , p6 , l1 , l2 , k ) (13)

Differentiating (13), we obtain the robot Jacobian J( p1 , p2 , p3 , p4 , p5 , p6 , l1 , l2 , k), where

each element is
∂xi
Jij = (14)
∂p j

4.3. Integral Control Using Synthetic Jacobian

The hybrid controller (7) was initially implemented using the synthetic Jacobian and
no proportional term (k p = 0). The implementation was equivalent to that in Section 4.1
with the difference being that the Jacobian is synthetic and continuously evaluated at
the current robot state xc . The setpoint was selected to be x = −5 mm, y = −0.5 mm,
z = 0 mm, θ = 6◦ , which is a representative point inside the workspace not close to
saturation. A value of k i = 0.3 was used. As in the previous Section 4.1, the value of k i was
selected experimentally using a heuristic process. Larger values of k i improve performance
but can lead to overshoot and oscillations, with the risk of reaching saturation, whereas
smaller values of k i yield a smoother response but reduce performance in terms of response
time. A compromise value of k i = 0.3 was used, which is higher than in Section 4.1 since
the control using the synthetic Jacobian showed lower oscillations.
Actuators 2024, 13, 242 15 of 20

The results are shown in Figure 7. The robot pose tends to the setpoint. The integral
action creates minor fluctuations in the robot, and the robot does not exactly reach the target,
both of which are due to imperfections in the synthetic kinematics model. Nonetheless,
the robot is within 0.5 mm of the target. It should be noted that the position error of the
electromagnetic tracker used has a standard deviation of 0.7 mm. The error is thus below
one standard deviation of the equipment error.
The hybrid control using a synthetic Jacobian performs well as long as there is no
saturation as illustrated in Figure 7.

Figure 7. Experimental results of setpoint regulation using a hybrid control combining IK NN and an
integral action based on a synthetic Jacobian.
Actuators 2024, 13, 242 16 of 20

4.4. Proportional Integral Control Using Synthetic Jacobian

The addition of the proportional term in the hybrid control is explored to mitigate the
issues described in the previous paragraph. The proportional term is intended to improve
the convergence speed and help prevent the controller from getting stuck when reaching
transient saturation states. It should be noted that there are two types of saturation:
transient and permanent. Transient saturation occurs during oscillations or transient
movements, where the robot unintendedly crosses into a configuration with saturation in
the pressure inputs. In these cases, the integral term stops updating, and the proportional
term can help the control recover. The aim for the proportional term is to provide a
contribution to the control that is continuously updated in a way that is proportional to the
error while the integral term is not updating, bringing the robot out of transient saturation.
Once saturation is no longer present, the integral term can be updated again.
The controller (7) was implemented in an equivalent manner as in the previous
subsection. The setpoint was selected to be x = −3 mm, y = −0.3 mm, z = 2 mm,
θ = 6◦ , which is also a point near saturation. The controller gains were selected as k i = 0.3
and k p = 0.02. The value of k i was maintained from the previous subsection given the
similarities in the control law. As in Section 4.3, k p was selected experimentally, exploring a
range values that kept the ratio f k p /k i similar to our previous work [13]. Larger values of
k p lead to faster response but also yield oscillations and overshoot. These can lead the robot
to a configuration with saturation, where it cannot recover. As a compromise, k p = 0.02
was selected.
The results are shown in Figure 8. As can be seen, the robot tends to the desired
setpoint. The robot displays a trade-off in the four tracked variables, which is due to model
inaccuracies, which means that even though the robot has four inputs, it cannot exactly
converge in all four outputs. Nonetheless, the robot reaches a submillimetre error. In
addition, the robot is capable of avoiding a region near saturation at approximately 1.8 s,
and recovers to tend to the setpoint.
Permanent saturation is a more complex problem that is not addressed in this work. It
occurs when the robot reaches a steady configuration with saturation, e.g., because to reach
the target, the robot needs to pass through an intermediate configuration where saturation
occurs for an extended period of time. The control methods proposed in this work fail in
those cases, leaving the robot stuck. Motion planning can be used to avoid regions with
permanent saturation, and this will be explored in future work.

4.5. Hybrid Control Discussion

Overall, we found the best hybrid control to consist of the IK NN for a first approxi-
mation of the required pressures combined with a PI control using a synthetic Jacobian to
converge to the desired setpoint. This control performs well as long as there is no saturation.
The robustness of this control, as well as that of the control formulation using an
experimental Jacobian, depends on the operation conditions. The control formulation
using the synthetic Jacobian is particularly vulnerable to model inaccuracies such as those
caused by changes in robot behaviour due to significant manufacturing tolerances, as well
as hysteresis and nonlinearities. If inaccuracies are very significant, which can occur in
extreme cases where manufacturing variations are coupled with robot nonlinearities, the
control can diverge because the Jacobian based on the model is not representative of the
robot behaviour. Even in cases where inaccuracies are smaller, they can reduce performance
since the synthetic model that underpins the control will not be accurate. As such, if a
high-density dataset is available, the hybrid control based on an experimental Jacobian
estimate (7) can perform better.
The experimental Jacobian-based control, however, has significant drawbacks as noted
in Section 4.1. The Jacobian can be distorted, and it changes abruptly between points,
particularly when datapoints are sparse, which means that it needs to be evaluated at
the target pose, as implemented in this work. The consequence is that, if the robot pose
moves far from the target, the Jacobian can cease to be representative of robot behaviour
Actuators 2024, 13, 242 17 of 20

and the control can fail. Thus, if the dataset is sparse, as in the case of our robot given
its limited lifespan, the synthetic Jacobian formulation is more generally suitable. The
synthetic Jacobian provides satisfactory experimental results.

Figure 8. Experimental results of setpoint regulation using a hybrid control combining IK NN and a
proportional–integral action based on a synthetic Jacobian.

The combination of both experimental and synthetic Jacobian-based control can be

advantageous by retaining the advantages of both. This could be achieved using transfer
learning to merge the experimental data of robot behaviour with the synthetic model.
However, this requires further research, and it will be addressed in future work.
Both synthetic and experimental Jacobian formulations offer some robustness to
unforeseen external disturbances, e.g., constant external loads. The integral part of the
control is designed to compensate for the effect of disturbances, and the proportional yields
faster adjustments. However, both formulations are vulnerable to disturbances that take the
robot to a configuration where the Jacobian is no longer an accurate representation of the
Actuators 2024, 13, 242 18 of 20

robot behaviour, e.g., in case the robot bends due to an interaction with a hard constraint
in the environment such as a wall. In those cases, the derivatives of pose with respect
to pressures in the Jacobian do not match reality. In general, the experimental Jacobian
is expected to be less robust to those unforeseen disturbances, given that it interpolates
between points in the training dataset. If the robot reaches a configuration outside the
training dataset, the experimental Jacobian is prone to failing. The synthetic Jacobian
instead can maintain a better representation of the robot behaviour even outside the training
dataset since it relies on the analytical model for extrapolation. Even so, if the robot is bent to
a great extent due to interaction with a hard constraint, the evaluation of the Jacobian based
on the current configuration relying on the model may not be representative of the robot
behaviour in practice, leading to large errors. The robustness of both control formulations
to changes in robot dynamics, e.g., those caused by a change in robot orientation relative
to gravity, is also expected to differ if the changes in dynamics lead to a configuration
where the Jacobian is less representative. For example, if the robot experiences a significant
change in bending angle due to gravity, neither of the Jacobian estimates may be accurate,
and the control may yield large errors. As in the previous case, the experimental Jacobian
is expected to be less robust in robot configurations outside the training dataset, while
the synthetic Jacobian is expected to extrapolate better, provided that the changes in robot
dynamics are relatively small.
The hybrid control with the synthetic Jacobian proposed in this work is an improve-
ment with respect to the control in [30]. The control in this work can operate in the entire
workspace rather than only in one quadrant. Moreover, this work is applicable to robots
with six inputs and four pose outputs, while [30] is limited to four inputs and three outputs.
In addition, the use of the Jacobian in this work is an advancement with respect to only
using the sign in the relation between inputs and outputs, which leads to better perfor-
mance. The hybrid control is also an advancement with respect to the hybrid control in [26]
since it applies to a two-segment robot. The hybrid control is also advantageous relative
to [13] since it works in the full workspace of a two-segment robot, and the NN captures
the behaviour of each prototype. The disadvantage relative to [13] is that the hybrid control
requires over 1000 datapoints for training, it is slower to converge, and it is less able to
compensate for disturbances.
Comparing the present hybrid control with ML and RL solutions in the literature, the
present control requires significantly less training data. Leading works for soft manipulators
such as [23] for RL and [20] for ML used in the order of 10,000 or more datapoints, whereas
in this work, we use 1369. The absolute errors in this work are also lower than in [23], and
the relative errors are comparable or lower. However, it should be noted that this present
work only focuses on setpoint regulation, whereas [23] addressed trajectory tracking. The
errors in [20,24] are lower than in this paper, but it should be noted that those works
were conducted on tendon-driven soft robots, which are typically simpler to model, more
repeatable, and with lower nonlinearities and hysteresis. The RL and ML approaches
in [20,24] have not yet been applied to the soft robot used in this paper, as they require a
robot with a significantly longer lifespan.

5. Conclusions
The control of a soft robotic manipulator was explored in this work using a hybrid
approach that combines ML to learn the robot kinematics and closed-loop control to
compensate for errors. Various ML approaches were explored to solve the DK and IK
problems for a manipulator with six pressure inputs and four robot pose outputs with
conditions on the combinations of pressure inputs. The most suitable DK NN and IK NN
were selected, and this was validated experimentally. A hybrid control formulation based
on an integral action was proposed, and two main approaches were explored to map the
pressure inputs to the pose output errors using the robot Jacobian. A Jacobian estimate
built using an experimental dataset was first developed, and the corresponding control
was tested. This showed correct performance as long as the IK NN bring the robot close to
Actuators 2024, 13, 242 19 of 20

the desired setpoint, but it showed the Jacobian estimate to be unreliable when deviations
from the setpoint are significant and when the Jacobain is distorted, typically near the
workspace edge. A second approach using a synthetic Jacobian developed assuming
piecewise constant curvature bending and using a robot model was then investigated.
Both integral control and PI control were built based on the synthetic Jacobian. This was
experimentally tested to show correct performance provided that there is no saturation.
The tests, however, highlighted the need for an improved control formulation when the
control saturates, and when the robot behaviour deviates from the model. Future work
will investigate improved models combining the theoretical model with experimental data
through physics-informed and transfer learning approaches.

Author Contributions: Conceptualisation, A.G.-C., F.S., V.F., E.F.; methodology, A.G.-C., F.S., V.F.;
software, A.G.-C., F.S., V.F.; validation, A.G.-C., F.S., V.F.; formal analysis, A.G.-C., F.S.; investigation,
A.G.-C., F.S., V.F.; resources, A.G.-C., F.S., V.F.; data curation, A.G.-C., F.S.; writing—original draft
preparation, A.G.-C., F.S.; writing—review and editing, V.F., E.F.; visualisation, A.G.-C., F.S., V.F.;
supervision, E.F.; project administration, A.G.-C., F.S.; funding acquisition, E.F. All authors have read
and agreed to the published version of the manuscript.
Funding: This work was in part supported by a collaboration with the Multi-scale Medical Robotics
Centre, Chinese University of Hong Kong. The funder was not involved in the study design, collection,
analysis, interpretation of data, the writing of this article or the decision to submit it for publication.
Fahim Shakib has been partially funded by the EPSRC grant Model Reduction from Data, Grant No.
EP/W005557.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Suzumori, K.; Iikura, S.; Tanaka, H. Development of flexible microactuator and its applications to robotic mechanisms. In
Proceedings of the 1991 IEEE International Conference on Robotics and Automation, Sacramento, CA, USA, 9–11 April 1991;
IEEE Computer Society: Washington, DC, USA, 1991; pp. 1622–1623.
2. Suzumori, K.; Iikura, S.; Tanaka, H. Applying a flexible microactuator to robotic mechanisms. IEEE Control Syst. Mag. 1992,
12, 21–27.
3. Cianchetti, M.; Nanayakkara, T.; Ranzani, T.; Gerboni, G.; Althoefer, K.; Dasgupta, P.; Menciassi, A. Soft Robotics Technologies
to Address Shortcomings in Today’s Minimally Invasive Surgery: The STIFF-FLOP Approach. Soft Robot. 2014, 1, 122–131.
[CrossRef]
4. Pagliarani, N.; Arleo, L.; Albini, S.; Cianchetti, M. Variable Stiffness Technologies for Soft Robotics: A Comparative Approach for
the STIFF-FLOP Manipulator. Actuators 2023, 12, 96. [CrossRef]
5. Chen, G.; Pham, M.T.; Maalej, T.; Fourati, H.; Moreau, R.; Sesmat, S. A Biomimetic steering robot for Minimally invasive surgery
application. In Advances in Robot Manipulators; IN-TECH: London, UK, 2009.
6. Dong, X.; Axinte, D.; Palmer, D.; Cobos, S.H.; Raffles, M.; Rabani, A.; Kell, J. Development of a slender continuum robotic system
for on-wing inspection/repair of gas turbine engines. Robot.-Comput.-Integr. Manuf. 2017, 44, 218–229. [CrossRef]
7. Zhang, J.; Li, Y.; Kan, Z.; Yuan, Q.; Rajabi, H.; Wu, Z.; Peng, H.; Wu, J. A Preprogrammable Continuum Robot Inspired by
Elephant Trunk for Dexterous Manipulation. Soft Robot. 2023, 10, 636–646. [CrossRef] [PubMed]
8. Kolachalama, S.; Lakshmanan, S. Continuum Robots for Manipulation Applications: A Survey. J. Robot. 2020, 2020, 4187048.
[CrossRef]
9. Treratanakulchai, S.; Franco, E.; Garriga-Casanovas, A.; Minghao, H.; Kassanos, P.; y Baena, F.R. Development of a 6 dof soft
robotic manipulator with integrated sensing skin. In Proceedings of the 2022 IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS), Kyoto, Japan, 23–27 October 2022; pp. 6944–6951.
10. Garriga-Casanovas, A.; Treratanakulchai, S.; Franco, E.; Zari, E.; Ferrandy, V.; Virdyawan, V.; y Baena, F.R. Optimised Design and
Performance Comparison of Soft Robotic Manipulators. In Proceedings of the 2022 7th International Conference on Mechanical
Engineering and Robotics Research (ICMERR), Krakow, Poland, 9–11 December 2022; pp. 129–136.
Actuators 2024, 13, 242 20 of 20

11. Ferrandy, V.; Ferryanto, F.; Sugiharto, A.; Franco, E.; Garriga-Casanovas, A.; Mahyuddin, A.I.; Rodriguez y Baena, F.; Mihradi,
S.; Virdyawan, V. Modeling of a two-degree-of-freedom fiber-reinforced soft pneumatic actuator. Robotica 2023, 41, 3608–3626.
[CrossRef]
12. Runge, G.; Wiese, M.; Raatz, A. FEM-based training of artificial neural networks for modular soft robots. In Proceedings of the
2017 IEEE International Conference on Robotics and Biomimetics (ROBIO), Macau, Macao, 5–8 December 2017; pp. 385–392.
13. Franco, E.; Garriga-Casanovas, A. Energy-shaping control of soft continuum manipulators with in-plane disturbances. Int. J.
Robot. Res. 2021, 40, 236–255. [CrossRef]
14. Roshanfar, M.; Dargahi, J.; Hooshiar, A. Cosserat Rod-Based Dynamic Modeling of a Hybrid-Actuated Soft Robot for Robot-
Assisted Cardiac Ablation. Actuators 2023, 13, 8. [CrossRef]
15. García-Samartín, J.F.; Rieker, A.; Barrientos, A. Design, Manufacturing, and Open-Loop Control of a Soft Pneumatic Arm.
Actuators 2024, 13, 36. [CrossRef]
16. Duan, J.; Zhang, K.; Qian, K.; Hao, J.; Zhang, Z.; Shi, C. An Operating Stiffness Controller for the Medical Continuum Robot
Based on Impedance Control. Cyborg Bionic Syst. 2024, 5, 0110. [CrossRef] [PubMed]
17. Zhang, S.; Li, F.; Fu, R.; Li, H.; Zou, S.; Ma, N.; Qu, S.; Li, J. A Versatile Continuum Gripping Robot with a Concealable Gripper.
Cyborg Bionic Syst. 2023, 4, 0003. [CrossRef] [PubMed]
18. Kim, D.; Kim, S.H.; Kim, T.; Kang, B.B.; Lee, M.; Park, W.; Ku, S.; Kim, D.; Kwon, J.; Lee, H.; et al. Review of machine learning
methods in soft robotics. PLoS ONE 2021, 16, e0246102. [CrossRef] [PubMed]
19. Lee, K.H.; Fu, D.K.; Leong, M.C.; Chow, M.; Fu, H.C.; Althoefer, K.; Sze, K.Y.; Yeung, C.K.; Kwok, K.W. Nonparametric online
learning control for soft continuum robot: An enabling technique for effective endoscopic navigation. Soft Robot. 2017, 4, 324–337.
[CrossRef] [PubMed]
20. Thuruthel, T.G.; Falotico, E.; Renda, F.; Laschi, C. Learning dynamic models for open loop predictive control of soft robotic
manipulators. Bioinspiration Biomimetics 2017, 12, 066003. [CrossRef]
21. Chin, K.; Hellebrekers, T.; Majidi, C. Machine Learning for Soft Robotic Sensing and Control. Adv. Intell. Syst. 2020, 2, 1900171.
[CrossRef]
22. Hyatt, P.; Wingate, D.; Killpack, M.D. Model-based control of soft actuators using learned non-linear discrete-time models. Front.
Robot. AI 2019, 6, 22. [CrossRef] [PubMed]
23. Centurelli, A.; Arleo, L.; Rizzo, A.; Tolu, S.; Laschi, C.; Falotico, E. Closed-loop Dynamic Control of a Soft Manipulator using
Deep Reinforcement Learning. IEEE Robot. Autom. Lett. 2022, 7, 4741–4748. [CrossRef]
24. Thuruthel, T.G.; Falotico, E.; Renda, F.; Laschi, C. Model-Based Reinforcement Learning for Closed-Loop Dynamic Control of Soft
Robotic Manipulators. IEEE Trans. Robot. 2019, 35, 124–134. [CrossRef]
25. Zhang, H.; Cao, R.; Zilberstein, S.; Wu, F.; Chen, X. Toward Effective Soft Robot Control via Reinforcement Learning. In Intelligent
Robotics and Applications, Proceedings of the 10th International Conference, ICIRA 2017, Wuhan, China, 16–18 August 2017; Springer
International Publishing: Heidelberg, Germany, 2017; pp. 173–184. [CrossRef]
26. Dai, Y.; Deng, Z.; Wang, X.; Yuan, H. A Hybrid Controller for a Soft Pneumatic Manipulator Based on Model Predictive Control
and Iterative Learning Control. Sensors 2023, 23, 1272. [CrossRef]
27. Zhang, J.; Liu, L.; Xiang, P.; Fang, Q.; Nie, X.; Ma, H.; Hu, J.; Xiong, R.; Wang, Y.; Lu, H. AI co-pilot bronchoscope robot. Nat.
Commun. 2024, 15, 241. [CrossRef]
28. Schölkopf, B.; Smola, A.J. Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond; MIT Press:
Cambridge, MA, USA, 2001.
29. Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning; MIT Press: Cambridge, MA, USA, 2016.
30. Treratanakulchai, S.; Franco, E.; y Baena, F.R. Model-free Position Control of a Soft Continuum Manipulator in Cartesian Space.
In Proceedings of the 2023 International Conference on Control, Automation and Diagnosis (ICCAD), Rome, Italy, 10–12 May
2023; pp. 1–6.
31. Garriga-Casanovas, A.; Collison, I.; Rodriguez y Baena, F. Toward a common framework for the design of soft robotic manipulators
with fluidic actuation. Soft Robot. 2018, 5, 622–649. [CrossRef]
32. Williams, C.K.; Rasmussen, C.E. Gaussian Processes for Machine Learning; MIT Press: Cambridge, MA, USA, 2006; Volume 2.
33. Rasmussen, C.E.; Nickisch, H. Gaussian processes for machine learning (GPML) toolbox. J. Mach. Learn. Res. 2010, 11, 3011–3015.
34. Astrom, K.J.; Rundqwist, L. Integrator windup and how to avoid it. In Proceedings of the 1989 American Control Conference,
Pittsburgh, PA, USA, 21–23 June 1989; pp. 1693–1698.
35. Garriga-Casanovas, A.; Rodriguez y Baena, F. Kinematics of continuum robots with constant curvature bending and extension
capabilities. J. Mech. Robot. 2019, 11, 011010. [CrossRef]
36. Murray, R.M.; Li, Z.; Sastry, S.S. A Mathematical Introduction to Robotic Manipulation; CRC Press: Boca Raton, FL, USA, 1994.
[CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.