Gear Wheels based Simulation of Crawlers for Mobile Robot Servosila

Engineer

Ruslan Gabdrahmanov

1 a

, Tatyana Tsoy

1 b

, Yang Bai

2 c

, Mikhail Svinin

2 d

and Evgeni Magid

1,3 e

Intelligent Robotics Department, Kazan Federal University, 35 Kremlin Street, Kazan, Russian Federation

College of Information Science and Engineering, Ritsumeikan University, Biwako-Kusatsu Campus 1-1-1 Nojihigashi,

Kusatsu, Japan

HSE University, 34 Tallinskaya Street, Moscow, Russian Federation

Keywords:

Rescue Robot, Crawler Robot, ROS, Modelling, Gazebo.

Abstract:

In a process of research, it is beneﬁcial to test new theories and early stage developments in virtual worlds of

an adequate realistic simulation before starting real world experiments. While modelling of wheeled mobile

robots is well-studied and typically does not imply signiﬁcant difﬁculties, a realistic modelling of a crawler

robot is a complicated task. This paper discusses several existing approaches for a crawler robot modelling in

Gazebo simulator and presents a new approach, which approximates each crawler with a set of gear wheels.

We compared several approaches for Servosila Engineer crawler robot modelling in Gazebo by their climb-

ing capabilities, velocity, acceleration and real time factor parameters with regard to the real robot. The

comparison results demonstrated that the new approach is feasible in terms of CPU load and provides a better

approximation to the real robot performance. Moreover, it successfully eliminated an issue of a crawler seizure

while climbing sharp edges of obstacles, which is typical for pseudo-wheels based approaches.

1 INTRODUCTION

An unmanned ground vehicle (UGV) is a mobile

robot of any type that moves through its operational

environment along various types of support surfaces,

which might range from a ﬂat surface to a rough

debris-like terrain. A UGV is a most widely used type

of a robot for a variety of real world tasks.

A crawler (or tracked) robot is a sub-type of mo-

bile robots that use different types of tracks as run-

ning gear. Crawler robots are employed when a task

requires an extended mobility of a vehicle, includ-

ing such tasks as planetary exploration, mining and

urban search and rescue (USAR). Typically, crawler

robots have higher level of manoeuvrability relatively

to wheeled robots of similar size and power, but are

inferior in terms of velocity and energy efﬁciency on

a ﬂat surface.

https://orcid.org/0000-0001-9276-2034

https://orcid.org/0000-0002-5715-7768

https://orcid.org/0000-0003-1080-1939

https://orcid.org/0000-0003-2459-2250

https://orcid.org/0000-0001-7316-5664

An urban search and rescue (USAR) robotics is a

most obvious example of an application that demands

a high level of a robot manoeuvrability.

USAR was introduced at the end of the 20th cen-

tury as a separate branch of a ﬁeld robotics, which

concentrates on mechanics of rescue robots, their nav-

igation, mapping and other classic tasks of robotics as

well as interaction of a human with a robot, all being

viewed through a prism of rescue related tasks. One

of the main tasks of USAR robotics is to search for

victims in partially damaged or completely destroyed

man-made structures. Therefore, a typical USAR task

environment for a UGV contains piles of trash and

debris formed by building materials, furniture, vari-

ous appliances, household and ofﬁce items that make

it difﬁcult to observe, localize and map the environ-

ment (Saﬁn et al., 2021; Malov et al., 2019).

Simulations are widely used in many areas of

robotics, including design and construction of a robot,

control algorithms development, education, demon-

strations, etc. (Yakovlev et al., 2015). They allow

creating virtual models of robots and a broad variety

of environments (Sheh et al., 2014; Simakov et al.,

2019), while targeting to achieve a high level of real-

Gabdrahmanov, R., Tsoy, T., Bai, Y., Svinin, M. and Magid, E.

Gear Wheels based Simulation of Crawlers for Mobile Robot Servosila Engineer.

DOI: 10.5220/0011355200003271

In Proceedings of the 19th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2022), pages 565-572

ISBN: 978-989-758-585-2; ISSN: 2184-2809

565

ism in behavior of these virtual models (Borisov et al.,

2016). Simulations in robotics are employed for many

reasons, including a typically very high cost of a real

world errors relatively to virtual world errors (Sha-

balina et al., 2019) and the ease of reproducing test

cases (Timperley et al., 2018).

Gazebo is a simulator that was designed specif-

ically for simulating robots and their environ-

ment (Foundation, 2021). Gazebo is fully compat-

ible with Robot Operating System (ROS) (Quigley

et al., 2009), allowing to create and use robot models

with ROS-based control systems without additional

labor. Other popular robotics simulators include We-

bots (Prabhakar et al., 2020), VRep (recently known

as CoppeliaSim (Yumbla et al., 2020)) and others. We

selected the Gazebo simulator because of its com-

patibility with ROS and high-quality physics, which

are the most important characteristics for our task of

a highly manoeuvrable crawler robot modelling and

model validation. There are several tools available

in Gazebo that are suitable for the implementation of

the running gear of robots, however, most of them are

intended for simulating wheeled or walking robots.

Since there is no ready-made template solution for

simulating active tracks, several different home-made

tool-based approximations were previously used by

researchers to simulate tracks of a crawler robot while

employing conventional wheeled robot approaches.

In this paper we present a new method of mod-

elling a crawler using geared pseudo wheels. The

approach demonstrated a stable behaviour and an ac-

ceptable performance in terms of a CPU and memory

load, which were indirectly measured via real time

factor (RTF) of Gazebo simulator while using RTF=1

for the real robot as a base benchmark. Moreover,

the resulting model performance parameters became

close to the real Servosila Engineer robot parameters

relatively to its previous models.

The rest of the paper is organized as follows. Sec-

tion 2 overviews existing solutions, including the ones

that were previously proposed by out team. Section 3

describes our new approach for the gear wheels based

simulation of crawlers and Section 4 presents a com-

parison of the real robot and four different models.

Finally, we conclude in Section 5.

2 EXISTING SOLUTIONS

At the time of writing this paper, there exists a number

of methods to simulate crawler robots’ chassis. These

methods could be divided into three main groups: a

brute-force generic acceleration, a wheel-based ap-

proach and a segmental tracks approach. Each ap-

proach has particular advantages and disadvantages,

which are described in this section.

2.1 Brute-force Generic Acceleration

A brute acceleration force is applied directly to a

robot chassis or to robot tracks, while a mobile part

of the chassis (tracks) is modelled with skid runners

or passive ﬁxed wheels. This method is the most pro-

ductive in terms of CPU and memory use (and thus

RTF), and is easy to implement. Yet, its correspon-

dence to a real physical robot model is questionable.

This approach could allow to achieve a relatively real-

istic behavior of a robot model solely on a ﬂat surface,

but in other cases the behavior of the model does not

correspond to the real robot (Pecka et al., 2017).

2.2 Wheel-based Approaches

This approach uses standard ﬁxed wheels (Shabalina

et al., 2018) in order to approximate a robot track. The

wheels could be visible when they completely replace

a rubber or a metal track, or they could be presented

in a virtual form. In the virtual form, the user sees a

visualization of a track while wheels, often referred

as pseudo-wheels, are responsible for physical inter-

action with environment. In both cases, wheels are

arranged in such a way that allows repeating a shape

of real robot tracks. Moreover, wheels could partially

overlap with each other while the physics of such col-

lisions between the wheels is ignored. We distinguish

two most popular wheel-based approaches as a single

line of large ﬁxed wheels (further refereed as a line-

of-wheels) and a large number of ﬁxed wheels that are

distributed along a track (further refereed as a circum-

ferential wheels).

Line-of-wheels approach (LWA) forms a struc-

ture where all wheels are located on a single straight

line. These are standard ﬁxed wheels of a diame-

ter, which is equal to a height of a track. Wheels

could virtually overlap each other without consider-

ing the inter-wheel collision physics (Figure 1) or

could be arranged without physical intersections (Fig-

ure 2, (Pecka et al., 2017)).

LWAs, especially with a small number of large

size wheels, look attractive in terms of simulation

performance (CPU, memory, RTF) and development

complexity. Yet, these approaches are suitable only

for a ﬂat surface, since a robot can unnaturally get

stuck while climbing an obstacle if a sharp part of the

obstacle gets between wheels (Sokolov et al., 2016)

(Figure 4, 5). Increasing a number of intersecting

wheels partially solves the problem (Pecka et al.,

2017), however, more wheels inevitably reduce per-

ICINCO 2022 - 19th International Conference on Informatics in Control, Automation and Robotics

566

Figure 1: Line-of-wheels approach with large virtual self-

intersecting wheels (one of optional models constructed by

the authors).

Figure 2: Line-of-wheels approach with large real non-

intersecting wheels (Pecka et al., 2017).

Figure 3: Line-of-wheels approach with large vir-

tual non-intersecting wheels with broad spaces between

wheels (Sokolov et al., 2016).

formance, while the robot could still get stuck in a

similar way, especially if it has a large mass.

Figure 4: LWA: A robot with large wheels gets stuck on a

sharp corner of an obstacle.

Figure 5: CWA: a robot with small wheels gets stuck on a

sharp corner of an obstacle.

Circumferential wheels approach (CWA) em-

ploys a large number of wheels that are placed along

a perimeter of each track. The wheels have typi-

cally equal radius, which is signiﬁcantly smaller than

a height of tracks (Figure 6, 7). Such implementation

achieves greater similarity in terms of geometry with-

out intersecting wheels, but suffers from poor simu-

lation performance associated with a large number of

wheels. It is also worth mentioning hybrid versions,

which use wheels of varying sizes in order to simulate

a track (Figure 8).

Figure 6: CWA: tightly packed small non-intersecting

wheels.

Figure 7: CWA: small wheels with spaces in-between

(Moskvin and Lavrenov, 2020).

2.3 Segmental Tracks

In this approach a track is assembled from rectangular

or more complex shape segments, which are linked

into a single chain by passive connections. he seg-

ments are stretched between at least two active rollers

Gear Wheels based Simulation of Crawlers for Mobile Robot Servosila Engineer

567

(a) The real robot.

(b) A hybrid model of the robot in USARSim.

Figure 8: Talon crawler robot and it’s model in USAR-

Sim (Pepper et al., 2007).

from front and rear edges of a chassis and are held

there using simulation physics or programmatically

(Figure 9, 10). Additional passive or active rollers

could be employed. Theoretically, such approach is

the most reliable since it has an almost absolute geo-

metric similarity, but, as a rule, it has a low produc-

tivity (Morita et al., 2018).

Considering the three approaches descried in this

section, segmental tracks seem to be the most attrac-

tive for developing a crawler model in terms of sim-

ilarity with a real robot since geometrically such im-

plementation is the closest one to real tracks. How-

ever, low computational efﬁciency and low stabil-

ity of this method make it unsatisfactory in prac-

tice (Sokolov et al., 2017; Kenwright and Morgan,

2012). Moreover, a robot with segmental tracks can

also get stuck if a sufﬁciently sharp or small obstacle

gets into a gap between track’s segments

Figure 9: Dynamic segmental tracks (Morita et al., 2018).

Figure 10: Another implementation of segmantal

tracks (Sokolov et al., 2017).

3 GEAR WHEELS BASED

SIMULATION

We propose a new approach of a track modelling,

which employs virtual gear wheels. Using such mod-

iﬁcation for LWA with large wheels could drastically

solve the problem of getting stuck without sacriﬁcing

computational performance. We added lugs to a stan-

dard ﬁxed wheel model, which are similar to those

found on most real tracks. We refer such wheel as a

gear wheel although it does not actually repeat a shape

of a real gear (Figure 11).

As a starting point, we used a model of Servosila

Engineer robot that was previously created by our

team (Moskvin and Lavrenov, 2020) using CWA with

small wheels, referred as pseudo-wheels. This model

is demonstrated in Figure 7: the white circles are the

pseudo-wheels (which are standard ﬁxed wheels) and

the black track is just a texture that forms a visual rub-

ber track without any physics behind it. The actual

physics of interaction with an environment (support-

ing plane) is delegated to the pseudo-wheels, which

could be switched on/off for visualization. Next, all

small pseudo-wheels were replaced by the new gear

ICINCO 2022 - 19th International Conference on Informatics in Control, Automation and Robotics

568

Figure 11: Single gear wheel.

wheels with of approximately the same size (Figure

12).

For virtual testing of the new track model con-

cept as well as for further comparative testing with

other models we employed a random step environ-

ment (RSE), or random stepﬁeld, which provides a

good approximation of an uneven terrain (Jacoff et al.,

2008). The new model achieved a declared by the

manufacturer passability and got rid of the issue of

seizure while climbing sharp edges of obstacles.

Unfortunately, the obtained RTF of the new

model is not acceptable for a comfortable use of the

model (Abbyasov et al., 2020). In the attempt to im-

prove the model performance in the terms of RTF, we

constructed another model that uses LWA with large

gear wheels, which have a similar size and shape of

protrusions to those of the real robot (Figure 13).

Figure 12: Servosila Engineer robot model with CWA and

small gear wheels.

Figure 13: Servosila Engineer robot model with LWA and

large gear wheels.

Initial angular position of each wheel is calculated

using the following relationship:

x = θ/3 ∗ n (1)

where θ is an angle between two nearest segments

drawn from a center of a wheel to a most distant (from

the wheel center) point of a protrusion on the wheel

(Figure 14); n is a counting number of a wheel - the

counting starts from n=1 for a last (rearmost) wheel

of each side of a simulated track.

Figure 14: Angle θ for an initial position calculation.

This approach allows avoiding an excessive un-

natural shaking of the robot in motion due to a syn-

chronous rotation of the non-round wheels. The ra-

dius of a wheel for odometrical data processing is

set as a radius of a circumscribed circle for the gear

wheel.

Gear wheels allowed to (almost completely) solve

the problem of a crawler seizure while climbing sharp

edges of obstacles that is caused by standard ﬁxed

wheels. The wheels protrusions allow a robot model,

similarly a real crawler robot, to ”cling” to an obsta-

cle surface at climbing and to achieve manoeuvrabil-

ity characteristics that closely correspond to the real

robot. However, due to their shape, the protrusions

complicate odometry and create a robot shaking ef-

fect while moving on a ﬂat surface. Nevertheless,

the shaking effect corresponds to a real crawler robot

shaking, which brings the model behavior closer to

the real robot.

The LWA large wheels implementation also

showed a sufﬁcient manoeuvrability and climbing

abilities without seizure. Comparatively to the CWA

small wheels model, the RTF was signiﬁcantly im-

proved achieving 0.6 for static cases and 0.5 for

dynamic cases when the new robot model travels

through a RSE. Moreover, the LWA large wheels

model is featured with a high degree of geometry sim-

ilarity with the real robot.

Gear Wheels based Simulation of Crawlers for Mobile Robot Servosila Engineer

569

4 COMPARISON OF MODELS

For virtual testing a number of typical for urban

search and rescue environments were constructed in

a form of a random step environment and four differ-

ent models of Servosila Engineer were validated with

the same settings in a teleoperational mode.

Table 1: Real robot dynamic characteristics.

Characteristic Real

robot

Declared by

the manu-

facturer

Max. linear velocity 0,4 m/s

1,39 m/s

Max. rotation velocity 8,5 rpm -

Max. elevation angle 45° 35°

Stopping distance with

max. linear velocity

1 cm -

We assume the velocity is limited by a low-level con-

troller.

4.1 Virtual Testing

Virtual tests of the models were carried out using typ-

ical obstacles of RSE, 20 runs per each robot and en-

vironment. The results are presented in Table 2.

Table 2: Obstacle tests.

Obstacle

type

Percentage of successful runs (%)

CWA

model

LWA

model

Model with

gear wheels

Random

RSE

0 15% 100%

Horizontal

barrier

0 10% 100%

Diagonal

barrier

0 0 40%

(a) CWA standard ﬁxed

wheels.

(b) CWA gear wheels.

Figure 15: Robot models on a randomly generated RSE.

The dynamic characteristics and passable obstacle

height of the new and previous robot models, and the

real robot were measured with virtual and real envi-

ronment tests that were performed in a teleoperational

mode (Figure 18 a, b; Figure 19 a, b). The results are

presented in Table 3.

(a) CWA standard ﬁxed

wheels.

(b) CWA gear wheels.

Figure 16: Robot models on RSE with a horizontal barrier.

(a) CWA standard ﬁxed

wheels.

(b) CWA gear wheels.

Figure 17: Robot models on RSE with diagonal barrier.

(a) Real robot overcomes 20

cm barrier (begin).

(b) Real robot overcomes 20

cm barrier (end).

Figure 18: Real robot on RSE with horizontal barrier.

(a) Robot model with CWA

overcomes 10 cm barrier.

(b) Robot model with gear

LWA overcomes 20 cm bar-

rier.

Figure 19: Robot models on RSE with horizontal barriers.

4.1.1 Results

The real robot parameters in Table 1 and performance

in RTF were used as a benchmark for the virtual tests,

which were described in the previous subsection. The

results of the tests are presented in Tables 2 and 3.

Virtual testing revealed the following qualitative

characteristics of the proposed gear wheel based ap-

proaches:

ICINCO 2022 - 19th International Conference on Informatics in Control, Automation and Robotics

570

Table 3: Comparison of models and real robot.

Model Max veloc-

ity

Max obstacle

height

Max angular

velocity

Max braking

distance

Acceleration RTF

Real robot 0,4 m/s

20 cm

8,5 rpm 0,02 m 0,2 m/s

1.0

CWA, small

standard

ﬁxed wheels

0,5 m/s 10 cm 4 rpm 1 m 0,1 m/s

0.2

LWA, large

standard

ﬁxed wheels

0,4 m/s 15 cm 2 rpm 1 m 0,1 m/s

0.65

CWA, small

gear wheels

0,4 m/s 20 cm

2 rpm 0,5 m 0,1 m/s

0.18

LWA, large

gear wheels

0,4 m/s 20 cm

4 rpm 0,3 m 0,2 m/s

0.65

A parallelepiped obstacle on a ﬂat surface. The robot moves without rocking or any other tricks that allow to overcome a

higher obstacle.

Presumably the velocity is limited programmatically on a low-level. In the robot speciﬁcations, the manufacturer announces

a maximal velocity of 5 km/h, which is 1,39 m/s

Technically, it is possible to overcome a height of up to 60 cm, but this requires to employ certain movement patterns and

balance control, which is very difﬁcult in a teleoperational mode.

• The models correspondence with dynamics and

geometry of the real robot is signiﬁcantly higher

than for any previously developed standard ﬁxed

wheels solutions. Yet, a model that could be

constructed using segmental method might allow

achieving even a better correspondence.

• The robot seizure problem while climbing sharp

edges of obstacles is eliminated.

• The performance in terms of Gazebo simulation

RTF is acceptable for LWA large gear wheels,

while the implementation for CWA small gear

wheels should be improved.

5 CONCLUSIONS

While modelling of wheeled mobile robots is well-

studied and typically does not imply signiﬁcant dif-

ﬁculties, a realistic modelling of a crawler robot is

a complicated task. This paper discussed several

existing approaches for a crawler robot modelling

in Gazebo simulator and presented a new approach,

which approximates each crawler with a set of gear

wheels. We compared several approaches for Ser-

vosila Engineer crawler robot modelling in Gazebo

by their climbing capabilities, velocity, acceleration

and real time factor parameters with regard to the real

robot. The real robot parameters and performance

were used as a benchmark for the tests. For virtual

testing a number of typical for urban search and res-

cue environments were constructed and four different

models of Servosila Engineer were validated with the

same settings in a teleoperational mode. The compar-

ison results demonstrated that two new gear wheels

based approaches are feasible in terms of CPU load

and provide a better approximation to the real robot

performance. Moreover, they successfully eliminated

an issue of a crawler seizure while climbing sharp

edges of obstacles, which is typical for pseudo-wheels

based approaches.

As a part of our future work we plan to take a

deeper look at the physics of the model relatively

to the real robot and to consider typical issues of a

crawler robot slipping and turning.

ACKNOWLEDGEMENTS

This paper has been supported by the Kazan Federal

University Strategic Academic Leadership Program

(”PRIORITY-2030”). The third and forth authors ac-

knowledge the support of the Japan Science and Tech-

nology Agency, the JST Strategic International Col-

laborative Research Program, Project No. 18065977.

REFERENCES

Abbyasov, B., Lavrenov, R., Zakiev, A., Yakovlev, K.,

Svinin, M., and Magid, E. (2020). Automatic tool for

gazebo world construction: from a grayscale image to

a 3d solid model. In 2020 IEEE International Con-

ference on Robotics and Automation (ICRA), pages

7226–7232. IEEE.

Borisov, A., Kuznetsov, S., Mamaev, I., and Tenenev, V.

(2016). Describing the motion of a body with an el-

liptical cross section in a viscous uncompressible ﬂuid

Gear Wheels based Simulation of Crawlers for Mobile Robot Servosila Engineer

571

by model equations reconstructed from data process-

ing. Technical Physics Letters, 42(9):886–890.

Foundation, O. S. R. (2021). Gazebo ofﬁcial site.

http://gazebosim.org/.

Jacoff, A., Downs, A., Virts, A., and Messina, E. (2008).

Stepﬁeld pallets: Repeatable terrain for evaluating

robot mobility. In Proceedings of the 8th Workshop

on Performance Metrics for Intelligent Systems, pages

29–34.

Kenwright, B. and Morgan, G. (2012). Practical introduc-

tion to rigid body linear complementary problem (lcp)

constraint solvers. In Algorithmic and Architectural

Gaming Design: Implementation and Development,

pages 159–201. IGI Global.

Malov, D., Edemskii, A., and Saveliev, A. (2019). Proactive

localization system as a part of a cyberphysical smart

environment. In 2019 International Conference on In-

dustrial Engineering, Applications and Manufactur-

ing (ICIEAM), pages 1–5. IEEE.

Morita, S., Hiramatsu, T., Niccolini, M., Argiolas, A., and

Ragaglia, M. (2018). Kinematic track modelling for

fast multiple body dynamics simulation of tracked

vehicle robot. In 2018 23rd International Confer-

ence on Methods & Models in Automation & Robotics

(MMAR), pages 910–915. IEEE.

Moskvin, I. and Lavrenov, R. (2020). Modeling tracks and

controller for servosila engineer robot. In Proceed-

ings of 14th International Conference on Electrome-

chanics and Robotics “Zavalishin’s Readings”, pages

411–422. Springer.

Pecka, M., Zimmermann, K., and Svoboda, T. (2017). Fast

simulation of vehicles with non-deformable tracks. In

2017 IEEE/RSJ International Conference on Intelli-

gent Robots and Systems (IROS), pages 6414–6419.

IEEE.

Pepper, C., Balakirsky, S., and Scrapper, C. (2007). Robot

simulation physics validation. In Proceedings of the

2007 Workshop on Performance Metrics for Intelli-

gent Systems, pages 97–104.

Prabhakar, M., Paulraj, V., Dhanraj, J. A., Nagarajan, S.,

Kannappan, D. A. K., and Hariharan, A. (2020). De-

sign and simulation of an automated guided vehicle

through webots for isolated covid-19 patients in hos-

pitals. In 2020 IEEE 4th Conference on Information &

Communication Technology (CICT), pages 1–5. IEEE.

Quigley, M., Conley, K., Gerkey, B., Faust, J., Foote, T.,

Leibs, J., Wheeler, R., Ng, A. Y., et al. (2009). Ros: an

open-source robot operating system. In ICRA work-

shop on open source software, volume 3, page 5.

Kobe, Japan.

Saﬁn, R., Lavrenov, R., and Mart

ınez-Garc

ıa, E. A. (2021).

Evaluation of visual slam methods in usar applica-

tions using ros/gazebo simulation. In Proceedings of

15th International Conference on Electromechanics

and Robotics” Zavalishin’s Readings”, pages 371–

382. Springer.

Shabalina, K., Sagitov, A., and Magid, E. (2018). Compar-

ative analysis of mobile robot wheels design. In 2018

11th International Conference on Developments in

esystems Engineering (dese), pages 175–179. IEEE.

Shabalina, K., Sagitov, A., Su, K.-L., Hsia, K.-H., and

Magid, E. (2019). Avrora unior car-like robot in

gazebo environment. In International Conference on

Artiﬁcial Life and Robotics, pages 116–119.

Sheh, R., Jacoff, A., Virts, A.-M., Kimura, T., Pellenz,

J., Schwertfeger, S., and Suthakorn, J. (2014). Ad-

vancing the state of urban search and rescue robotics

through the robocuprescue robot league competi-

tion. In Field and service robotics, pages 127–142.

Springer.

Simakov, N., Lavrenov, R., Zakiev, A., Saﬁn, R., and

Mart

ınez-Garc

ıa, E. A. (2019). Modeling usar maps

for the collection of information on the state of the en-

vironment. In 2019 12th International Conference on

Developments in eSystems Engineering (DeSE), pages

918–923. IEEE.

Sokolov, M., Afanasyev, I., Lavrenov, R., Sagitov, A.,

Sabirova, L., and Magid, E. (2017). Modelling

a crawler-type ugv for urban search and rescue in

gazebo environment. In Artiﬁcial Life and Robotics

(ICAROB 2017), International Conference on, pages

360–362.

Sokolov, M., Lavrenov, R., Gabdullin, A., Afanasyev, I.,

and Magid, E. (2016). 3d modelling and simulation of

a crawler robot in ros/gazebo. In Proceedings of the

4th International Conference on Control, Mechatron-

ics and Automation, pages 61–65.

Timperley, C. S., Afzal, A., Katz, D. S., Hernandez, J. M.,

and Le Goues, C. (2018). Crashing simulated planes is

cheap: Can simulation detect robotics bugs early? In

2018 IEEE 11th International Conference on Software

Testing, Veriﬁcation and Validation (ICST), pages

331–342. IEEE.

Yakovlev, K., Baskin, E., and Hramoin, I. (2015). Grid-

based angle-constrained path planning. In Joint Ger-

man/Austrian Conference on Artiﬁcial Intelligence

unstliche Intelligenz), pages 208–221. Springer.

Yumbla, F., Yumbla, E. Q., and Moon, H. (2020). The bi-

oloid gp robot with different conﬁgurations for simu-

lation in v-rep controlled by the robot operating sys-

tem (ros). In 2020 6th International Conference on

Control, Automation and Robotics (ICCAR), pages

54–58. IEEE.

ICINCO 2022 - 19th International Conference on Informatics in Control, Automation and Robotics

572