Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Path Planning and Obstacle Avoidance using ROS

1
Jayanth Kumar N., 2M U Abhisha, 3Adarsh, 4Uravakili Satheesh Reddy, 5Vidya Dudhanikar
5
Senior Assistant Professor
Electronics & Communication Engineering
Mangalore Institute of Technology & Engineering Moodabidri, Mangalore, India

Abstract:- The project presents an implementation of a II. ALGORITHMS

prototype for the autonomous vehicles with the Robot
Operating System (ROS). The system utilizes turtlebot 3 A. Gaussian Potential field algorithm
for the navigation purpose. The aim of this work is to Potential field-based methods use an artificial potential
integrate the obstacle avoiding algorithm with path field consisting of an attractive field and a repulsive field.
planning techniques to create a system that is able to The attractive field attracts the vehicle toward the goal while
navigate in given static environments. The software the repulsive field repels the vehicle from obstacles. Based
prototyping tool used is Robot Operating System. on the position vector of each obstacle the obstacles are
detected and avoided. Obstacle avoidance is one of the
Robots and their applications evolved swiftly over essential technologies in local path planning and one of the
decades, proving to be a great achievement in human critical technologies that guarantees human and vehicle
history. Robotics are improving people's lives by assisting safety.
them in resolving challenges they face on a daily basis. As
a result, they have a greater impact on our lives. The B. Dijkstra’s algorithm
influence of robot use on our lives is undeniably beneficial Dijkstra's algorithm is an algorithm we can use to find
in terms of saving time and effort. Robotics' constant shortest distances or minimum costs depending on what is
advancement has created several growth potentials in represented in a graph. Dijkstra’s algorithm is very similar
industries, education, utility facilities, and health care. to Prim’s minimum spanning tree. Like Prim’s Minimum
spanning tree, we generate a shortest path tree with a given
Keywords:- Turtlebot 3, ROS, Gazebo, RVi, Autorace. source as a root. We maintain two sets, one set contains
vertices included in the shortest-path tree, other set includes
I. INTRODUCTION vertices not yet included in the shortest-path tree. At every
step of the algorithm, we find a vertex that is in the other set
The Robot Operating System (ROS), a framework for and has a minimum distance from the source.
developing robot applications, allows programmers to piece
together a large system by connecting existing solutions to III. LITERATURE REVIEW
small problems. The main aspect of ROS is the way the
programme operates and interacts, which allows users to “Automatic Vehicle Navigation using Dijkstra’s
create complicated software without having to understand Algorithm”: Dijkstra's method is used to formulate the
how specific hardware works. ROS gives access to a network shortest path algorithm in this study. The classic Dijkstra
of processes (nodes) to communicate with a central hub. method can only discover the shortest path using the AS/RS
Nodes can run on many devices and communicate with the system's rectangular environment map, and it skips over
hub in a variety of ways. Doing this on a real robot would be additional pathways with the same distance. An updated
expensive, and it would waste time setting up the robot every Dijkstra algorithm is used to address the path-planning
time. As a result, robotic simulations are used. Gazebo is the problem in rectangular contexts, which identifies all
most commonly used ROS simulator. It has a strong equidistant shortest pathways[1]. This enables a car to drive
community, is open source, and is simpler to deploy robots itself around a network of interconnected roadways, taking
on. ROS was created with certain use cases in mind. Many the cheapest path.
things have changed since then, including a resurgence in AI
research and an increase in the number of use cases. Even “Functional System Architectures towards Fully
though ROS handles these issues admirably, robotics is Automated Driving” The paper covers how to determine the
growing more popular among the general public. vehicle's performance using the Functional System
Architecture. This study focuses on the creation of a system
In the near future, robots will continue to revolutionize architecture that is in line with the Robust Sense objectives
our lives beyond our wildest dreams, and they will continue across Europe. Rather than comparing individual algorithms
to play a significant role in human existence and behavior. It for perception, prediction, and planning, this paper
creates effective learning, production, control, and usage emphasizes robustness and reliability by focusing on
while also providing modern facilities for our lives. autonomous vehicle architectural design and its impact on
degraded operation modes in the case of sensor failures,
misdetections, and unexpected changes in traffic flow[2].
There are two sorts of architectures: centralized and
distributed architectures. Both topologies use redundant
sensor arrays as a shared feature.

IJISRT22JUN067 www.ijisrt.com 139

Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
“LOAM: Lidar Odometry and Mapping in Real-time” the path. As a result of this approach, the vehicle's geometry
In this article, range measurements from a 2-axis lidar and kinematics are incorporated into the path set, which is
travelling in 6-DOF are employed as a real-time approach for called a motion primitive and a rotation matrix is used to
odometry and mapping. Odometry algorithms estimate describe the novelty[7]. State lattice refers to the set of
velocity from lidar data, correct distortion in the point cloud, motion primitives arranged in a particular way. Motion
and then integrate and register the point cloud into a map primitives are saved in structural arrays and later used to
using a mapping technique[3]. The method is low-drift and construct a map's viable path. A path planning technique
computationally simple because it does not require high based on state grids is provided in general. When
precision range or inertial measurements. SLAM can be determining the path, a pre-computed path set also saves
performed by separating the problem into two halves and time. To validate the technique, different situations are
employing two algorithms to optimize a large number of considered.
parameters at the same time. One approach uses high-
frequency odometry with limited fidelity to determine the “Research on Obstacle Avoidance Path Planning
lidar's velocity. Algorithm for Six-axis Robot” In this paper, it deals with the
problem of the six-axis robot possibly colliding with an
“Path-planning of Automated Guided Vehicle based on obstacle in the working space[8]. This problem is addressed
Improved Dijkstra Algorithm” This paper introduces a new by using a method known as artificial potential fields.
algorithm that finds shortcuts in a rectangular area accessible Collision detection algorithm plans the manipulator's path to
to it all equal distance distances. By adding a run time be collision-free. A six-axis manipulator simulation system
planning test method, the right way for both short distance using MFC programming and OpenGL is developed to verify
and time can be determined[4]. The algorithm was developed the algorithm.
using Visual C ++, and simulation results show that it works
and is possible. Determining methods in Default Storage and “Static Obstacles Avoidance in Autonomous Ground
Recovery Programs (AS / RS). Using the adjacency matrix Vehicle using Fuzzy Logic Controller” For an autonomous
and spatial coordinates in two dimensions, the environment vehicle to be used in real time navigation technology is
map is described. Automated guided vehicles will be more probably an important requirement[9]. A fuzzy logic-based
efficient with this algorithm as it is applied to AS/RS. With controller is used in this paper to control an autonomous
the study, AS/RS system path planning will be supported, as vehicle for the challenging task of avoiding obstacles, which
well as other applications with multiple equidistant shortest is one of the soft computing techniques. Two inputs and one
paths. output are considered in the fuzzy logic controller. The
distances from left and right sensors are inputs, while the
“Autonomous Obstacle Avoidance Vehicle using rotation angle is an output. Fuzzy logic controllers with
LIDAR and an Embedded System” The car just has one Graphical User Interface [GUI] were designed based on 36
LIDAR sensor, allowing it to navigate in low-light conditions rule bases, and they provide information on the direction in
safely. Raspberry Pi and LIDAR sensor are the two main which the vehicle is moving.
hardware components. of this application[5]. A Light
Detection and Ranging module (LIDAR) were utilized to O. Khatib: This work provides a unique obstacle
detect obstacles. In LIDAR, pulses of pulsed light are avoidance strategy for manipulators and mobile robots based
continuously sent into every direction (360*) and measured on the concept of artificial potential fields[10]. In order to
with a sensor based on the wavelength and time at which the apply the potential field approach to robot manipulators, we
reflected pulses are received. The distance from the nearest use the operational space formulation. Formally, this
obstacle is determined by the wavelength and time of the formation focuses on controlling the motion of the end
signals received. It is powered by an OpenCV Raspberry Pi 3 effectors and avoiding obstacles in the operation space of the
board. Using current Wi-Fi infrastructure, the car is capable manipulator end effector. Its base tool is the equation of
of locating itself in the environment and safely driving from a motion of the manipulator end effectors in the operation
starting point to a destination point. space.

“A Study of Improved Global Path Planning Algorithm “Stereo vision obstacle avoidance using depth and
for Parking Robot Based on ROS” This paper focuses to elevation maps” The goal of this study is to present a single-
prepare a global path for the parking robots. The concept of sensor obstacle avoidance approach for mobile autonomous
global path planning is used for the parking purpose. This robots employing a stereo camera[11]. To avoid obstacles, a
method improved the efficiency of the robots that helps in robot must be able to determine the distance to its
parking[6]. Also, this algorithm was compared with other surrounding objects. Stereo vision is the most popular
algorithms on simulation platforms like ROS and found that method for extracting depth information from visual images,
this algorithm used has given good results than the others. but its disadvantage is that it requires very powerful
The concept of BP neural networks was also considered as hardware. A depth map concept uses images to align to
one of the main factors while testing. obtain information about the depth of an object. A elevation
map concept is used to take into account nearby objects to
“Path Planning Using State Lattice For Autonomous overcome the disparity map problem. As a result, the robot's
Vehicle” Several categories of planning for autonomous view is represented as a 3D point cloud. Threshold
vehicles are discussed in this paper, including path, Estimation Method is used for obstacle avoidance. To
trajectory, and motion. State lattice strategy is used to plan estimate the distance between objects, this algorithm uses

IJISRT22JUN067 www.ijisrt.com 140

Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Disparity maps robotically. ExaBot is used for this “Efficient Visual Obstacle Avoidance for Robotic
experiment. Mower Yao” This paper provides a novel technique of
robotic mower lawn obstacle avoidance and motion control
“UAV Obstacle Avoidance Using Potential Field under using the Gabor texture-based classification method and
Dynamic Environment” This paper implements an obstacle drivable regions search. After getting real-time image
avoidance approach that may be applied to the Parrot AR streams of grass scenes, they used Gabor filters to extract
Drone 2.0. Attractive potential functions and repulsive robust texture features. The robotic mower was tested under
potential functions are the two types of potential functions. complex scenes utilizing the Robot Operating System (ROS)
The quadcopter was attracted to the desired position by the built on compressed features[15]. Robotic lawnmowers have
attractive potential function and repelled the collision by the made remarkable advancements over the past decade thanks
repulsive potential function in order to keep it away from the to the massive demand. For security and efficiency reasons,
obstacles. They simulated quadcopter performance with static robotic mowers need to be capable of detecting and avoiding
impediments, dynamic obstacles from other quadcopters, and obstacles while operating. At present, collision sensors are
five quadcopters flying in the same field at the same time to mostly used to avoid obstacles, seasons, and light conditions
complete this study. A pair of quadcopters have been tested on robotic mowers.
with Krogh's repulsive potential. In this simulation, the
potential function of Krogh was investigated for dynamic “Design and Implementation of UAV Obstacle
obstacles[12]. Under ROS (Robot Operating System), the Avoidance System” An ultrasonic-based method of obstacle
Gazebo Simulator is used for simulation. They were able to avoidance is presented in this paper. The obstacle avoidance
simulate the Parrot AR Drone 2.0 as closely as possible using flight of the multi-rotor UAV can be achieved by increasing
the TUM simulator ROS module. To achieve the goal of the number of ultrasonic waves so that they form a circle
quadcopters, they have utilized the Khatib's attractive based on the traditional ultrasonic obstacle avoidance
potential feature. scheme, fusing the information from multiple ultrasonic
sensors, and adjusting the channel value of the tele controller
“Navigation of Autonomous Mobile Robots - Invited based on the distance between the UAV and the obstacle[16].
Paper” Through this paper we explored navigation stacks, At the same level, the motors are on the front, the back, the
mappings and localizations, sensor fusions, collision left, and the right side of the body. The four-rotor UAV must
avoidance and trajectory tracking[13]. Indoor and outdoor increase or decrease its power output simultaneously in order
navigation are the two different types of navigation for to avoid obstacles, so that the lift force of the UAV exceeds
autonomous vehicles. It is difficult for the navigation path to or decreases its speed. In order to achieve front-back and left-
be prepared for the unstructured environment. Using right motions on an aero plane in the horizontal plane, some
mapping and other concepts, we are able to overcome this force must be applied.
challenge. Maps are created by converting information in a
map's form. If the map is static, it will continue to have the “Fusing LIDAR sensor and RGB camera for object
same characteristics over time. There are most likely to be detection in autonomous vehicle with fuzzy logic approach”
dynamic maps that move with the vehicle as it moves. As a In this study, Velodyne HDL-64E 3D-LIDAR scanners and
result, the robot must construct the map of any moving RGB video cameras were used as the sensors. Lidar scanners
environment. Using local and global mapping techniques is acquire three- dimensional data of the environment in the
one way to accomplish this. When the robot has obtained form of a sparse point cloud[17]. Lidar sensors can be used
information about the environment around him, he can move for a variety of purposes, such as object detection, semantic
accordingly. It is also capable of localizing it segmentation, and obstacle detection. This study proposes a
method for fusing two sensors using fuzzy logic inference.
“Navigation and Obstacle Avoidance System in The camera image and the LIDAR point cloud have been
Unknown Environment” We learned about the concept of pre- processed prior to implementing our proposed
unmanned aerial vehicles in this work (UAVS), commonly framework, such as rectification and synchronization.
referred to as drones, that are very versatile, capable of
monitoring, rescuing and conducting searches from the skies. IV. PROBLEM STATEMENT
Commercial organizations, researchers, and individuals have
become interested in drones because of this, resulting in an Designing robots in real time is not always possible. It
increase in drones every day. As the number of airborne takes a lot of time and human effort. Moreover, it is very
vehicles increases, collisions may occur between them and much costly to implement the design in real time. This
navigation systems could be interfered with. Earlier studies problem can be overcome by means of using ROS which
of obstacle avoidance systems focused on path planning for allows the users to check for the possibilities of their robotic
static environments or known environments, but only a small system design.
number studied dynamic environments or unknown
environment[14]. By using lightweight LIDAR (Light Making use of this available resources we can design a
Detection and Ranging) sensors, the authors have devised an system design for the field of autonomous vehicles, ware
obstacle avoidance system that uses these sensors to detect house management, agriculture and so on. We know that a
obstacles and react accordingly. machine never gets bored with its tasks. Path planning and
obstacle avoidance approach can be used in automobile field
for the betterment of road transport.

IJISRT22JUN067 www.ijisrt.com 141

Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
V. METHODOLOGY VI. RESULTS

An autonomous car which is also known to us as Self Deployed the turtlebot3 and some obstacle into the
driving cars are becoming more popular these days. They are gazebo environment. Using the various options available in
designed in such a way that makes the work of the driver the Gazebo platform, we explored the ways in which the
very easy. This is made possible by considering all the static prototype could be implemented. We started exploring it
and dynamic environmental factors. We focussed on with the various options that were available with us.
developing a ROS prototype with, Gazebo simulator will be
used for simulating the bot in desired environment and the
Rviz visualizer will be used for visualizing the environment.
Firstly, the perception will be done which means collecting
the environment data from the sensors. Following perception
mapping will be done based on that bot will be localized with
respect to the framed map.

Path planning will lead to give ready to go environment

for bot Navigation stack will function with move base
package in this whole flow. The main purpose of this project
is to develop a prototype that can be used for the autonomous
vehicles and other fields like agriculture and ware house Fig 2: Gazebo simulator with Turtle Bot3 and obstacle
management for the supply of goods purposes.
The gazebo simulator provides various size of obstacles
which are of cylindrical, circular and others shapes. By
making use of this as obstacles for the turtlebot 3, we
designed the environments.

Fig 3: Hexagonal Gazebo environment with multiple

obstacles and Turtlebot3
As the obstacles and the turtlebot 3 are deployed into
Fig. 1: Block diagram the environment, we trained the Turtlebot3 in the gazebo
environment, by making it move frontwards, backwards, and
The Turtlebot3 is used as the bot for the purpose of this also on either side i.e. left and right.
project. There are two types of turtlebot3 available for usage
in ROS namely, the turtlebot3 burger and turtlebot3 waffle.
They got their names as they resemble, a burger and a waffle
respectively. Either of these can be used for this purpose. But
we are making use of the turtlebot3 waffle for the navigation
purpose. Firstly, the environment is created by using the
perception mapping, in which the turtle bot 3 is used as the
virtual bot for the simulation purpose. The starting and
destination places are set in the environment containing many
obstacles.

The Gaussian Potential Field algorithm is used for the

obstacle avoidance. The turtle Bot 3 will get trained using Fig. 4: Tutlebot3 moving in Gazebo environment
this algorithm. Also, for path planning, the Dijkstra’s
algorithm is used to find the shortest path that bot can make The RViz visualizer is visualizing the environment from
use to reach the destination. This entire simulation can be the perspective of the turtlebot3 waffle, as it moves along the
done using the navigation stack, which forms the base for this environment. Visualizing the map is done using RViz. As the
project along with the Robot Operating System Software. turtle Bot is made to move through the hexagonal gazebo
environment, the RViz will do the perception mapping for
the same. Once the bot completes exploring the complete
environment, the visualized map will look as shown below:

IJISRT22JUN067 www.ijisrt.com 142

Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

Fig 5: Completely visualized map

Fig 8: Developed an environment with classroom
The SLAM (Simultaneous Localization And Mapping) setup using Build Editor
generates the entire map simultaneously as the turtlebot3
moves around the space in the gazebo environment. VII. CONCLUSION

The Gazebo simulator provides various options for the

user to use as obstacles in any environments and also the
user can design their own environment and train the turtle
Bot 3 with that particular environment and as it visualizes the
complete environment, then we can make it move from
source point to destination without colliding with the
obstacles. By making the best use of the available platforms,
we can make sure that whether the systems can be
implemented properly for the benefit of human beings.

ACKNOWLEDGMENT
Fig 6: Visualized environment with mapping
We express our gratitude to our project guide Mrs.
As soon as the turtlebot 3 get trained and able to
Vidya Dudhnikar, Senior Assistant Professor, Department of
visualize the environment, we can train the model. We have
Electronics & Communication Engineering, Mangalore
trained the turtlebot 3 with Dijkstra’s algorithm for finding
Institute of Technology & Engineering, who always stood
the shortest path to navigate to the desired destination
behind us and supported in each and every step of this
without colliding with the obstacles.
project.

REFERENCES

[1.] M. Parulekar, V. Padte, T. Shah, K. Shroff, and R.

Shetty, “Automatic vehicle navigation using Dijkstra’s
Algorithm,” 2013 Int. Conf. Adv. Technol. Eng. ICATE
2013, 2013, doi: 10.1109/ICAdTE.2013.6524721.
[2.] Institute of Electrical and Electronics Engineers, 2016
IEEE Intelligent Vehicles Symposium (IV) : 19-22 June
2016. .
[3.] J. Zhang and S. Singh, “LOAM: Lidar Odometry and
Mapping in Real-time,” no. October 2015, 2015, doi:
Fig 7: Navigating through the shortestpath using Dijkstra’s 10.15607/rss.2014.x.007.
algorithm [4.] IEEE Control Systems Society, Dongbei da xue,
Chinese Association of Automation. Technical
By using the 2D navigation tool application given in the Committee on Control and Decision of Cyber Physical
Gazebo platform, we can set the destination for the turtlebot Systems, IEEE Singapore Section. Industrial
to move. without any collision. This 2D navigation tool also Electronics Chapter, Chongqing da xue, and Institute of
point the red color arrow mark as the destination is given and Electrical and Electronics Engineers., Proceedings of
the bot will move to the destination. Also, in this the turtlebot the the 29th Chinese Control and Decision Conference
moves in between the obstacles to reach the destination. The (2017CCDC) : 28-30 May 2017, Chongqing, China. .
obstacles are visualized in the color of green and the [5.] Aristoteleio Panepistēmio Thessalonikēs. Research
boundaries of the environment are in black color as shown in Dissemination Center and Institute of Electrical and
the figure 8. On similar basis, the turtlebot 3 is made to get Electronics Engineers, 2019 8th International
trained with the environment that is created and the Conference on Modern Circuits and Systems
classroom is made localized to the turtlebot 3 by moving it all Technologies (MOCAST) : May 13-15, 2019, Aristotle
around the new environment. University Research Dissemination Center (KEDEA),
Thessaloniki, Greece. .

IJISRT22JUN067 www.ijisrt.com 143

Volume 7, Issue 6, June – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[6.] L. Yan, L. Qi, K. Feiran, C. Guang, and C. Xinbo, “A
study of improved global path planning algorithm for
parking robot based on ROS,” 2020 4th CAA Int. Conf.
Veh. Control Intell. CVCI 2020, no. Cvci, pp. 607–612,
2020, doi: 10.1109/CVCI51460.2020.9338469.
[7.] S. Deb and S. Member, 2017 International Conference
on Technological Advancements in Power and Energy
(TAP Energy)., vol. 1. IEEE, 2017.
[8.] S. Jiang, H. Fang, K. He, and C. Yan, “Research on
obstacle avoidance path planning algorithm for six-axis
robot,” 2018 IEEE Int. Conf. Inf. Autom. ICIA 2018, no.
August, pp. 465–469, 2018, doi:
10.1109/ICInfA.2018.8812545.
[9.] M. Singh, S. Das, and S. K. Mishra, “Static obstacles
avoidance in autonomous ground vehicle using fuzzy
logic controller,” 2020 Int. Conf. Emerg. Technol.
INCET 2020, pp. 1–6, 2020, doi:
10.1109/INCET49848.2020.9154145.
[10.] O. Khatib, “Real-Time Obstacle Avoidance for
Manipulators and Mobile Robots,” Auton. Robot Veh.,
pp. 396–404, 1986, doi: 10.1007/978-1-4613-8997-
2_29.
[11.] T. Pire, P. De Cristóforis, M. Nitsche, and J. Jacobo
Berlles, “Stereo vision obstacle avoidance using depth
and elevation maps,” IEEE RAS Summer Sch. "Robot
Vis. Appl. VI Lat. Am. Summer Sch. Robot. Santiago,
Chile. December 3-7, 2012, pp. 1–3, 2012.
[12.] A. Budiyanto, A. Cahyadi, T. B. Adji, and O.
Wahyunggoro, “UAV obstacle avoidance using
potential field under dynamic environment,” ICCEREC
2015 - Int. Conf. Control. Electron. Renew. Energy
Commun., pp. 187–192, 2015, doi:
10.1109/ICCEREC.2015.7337041.
[13.] J. Z. Sasiadek, Y. Lu, and V. Polotski, “Navigation of
autonomous mobile robots - Invited paper,” Lect. Notes
Control Inf. Sci., vol. 360, no. May 2014, pp. 187–208,
2007, doi: 10.1007/978-1-84628-974-3_17.
[14.] J. Singh, M. Dhuheir, A. Refaey, A. Erbad, A.
Mohamed, and M. Guizani, “Navigation and Obstacle
Avoidance System in Unknown Environment,” Can.
Conf. Electr. Comput. Eng., vol. 2020-Augus, 2020,
doi: 10.1109/CCECE47787.2020.9255754.
[15.] Čhulālongkō̜̄ nmahāwitthayālai and Institute of
Electrical and Electronics Engineers, 2017 2nd
International Conference on Control and Robotics
Engineering (ICCRE2017) : April 1-3, 2017, Bangkok,
Thailand. .
[16.] Y. Xu, M. Zhu, Y. Xu, and M. Liu, “Design and
Implementation of UAV Obstacle Avoidance System,”
Proc. - 2019 2nd Int. Conf. Saf. Prod. Informatiz.
IICSPI 2019, pp. 275–278, 2019, doi:
10.1109/IICSPI48186.2019.9095902.
[17.] T. Nguyen and M. Yoo, “Fusing LIDAR sensor and
RGB camera for object detection in autonomous vehicle
with fuzzy logic approach,” Int. Conf. Inf. Netw., vol.
2021-Janua, pp. 788–791, 2021, doi:
10.1109/ICOIN50884.2021.9334015.

IJISRT22JUN067 www.ijisrt.com 144