Survey On Path Planning and Navigational Algorithms

Saurabh Pethkar, Anirudhh Kalkura, Krishna Reddy

Drexel University

Abstract - This paper comprehensively surveys various path based algorithms: the Wavefront Algorithm and
planning and navigational algorithms, highlighting their Voronoi Diagrams, elucidating their mechanisms,
unique characteristics, applications, and limitations. The
applications, and limitations.
focus is on grid-based, graph-based, sampling-based,
dynamic programming, heuristic, AI and ML-based, fuzzy
logic and expert systems, hybrid, and bio-inspired 1. Wavefront Algorithm
algorithms, and a special mention of bug algorithms. Each The Wavefront Algorithm is a pathfinding method
category is examined for its method, efficiency, and renowned for its simplicity and effectiveness in
suitability in different scenarios. The paper aims to offer structured environments. This algorithm functions
insights into how these algorithms are integral to robotics,
by assigning numerical values to the cells in a grid,
autonomous navigation, and virtual environment mapping.
Comparative analysis and discussions on the adaptability
starting from the goal point. The value assigned to
and robustness of these algorithms, particularly in dynamic each cell represents its distance from the goal, with
and complex environments, supply valuable perspectives for the goal cell typically assigned the value of zero.
future technological advancements. This process creates a gradient or 'wavefront'
across the grid. The algorithm systematically
Introduction propagates these values outward from the goal,
Path planning and navigation algorithms are pivotal incrementally increasing them with each step away
in shaping the future of autonomous systems. This from the goal.
paper delves into the intricacies of various When navigating, a robot or agent starts at its
algorithms that form the backbone of such systems. initial position and moves toward adjacent cells
We begin by exploring the fundamental concepts with lower values, effectively following the
underlying path planning and navigation, laying a gradient toward the goal. This deterministic
foundation for understanding the complexities approach ensures that if a path exists, the algorithm
will find it. However, it's most effective in static
involved. We then address the significance of these
environments, as dynamic changes require
algorithms in diverse applications, ranging from recalculating the entire grid. The algorithm's
robotics to autonomous vehicle navigation. The memory consumption can also be considerable,
aim is to supply a detailed overview, emphasizing especially in large environments.
the critical role these algorithms play in decision-
making processes and their impact on the
effectiveness of autonomous systems. This
introduction sets the stage for a comprehensive
exploration of several types of algorithms, each
with its unique method and application. A type of
many-valued logic known as fuzzy logic deals with
approximations in reasoning as opposed to precise
and fixed inferences. Because exact and binary
logic can't always account for the complexity and
unpredictability of real-world surroundings, it's
especially helpful in robot path planning and
navigation. Fuzzy logic is used in robotics to make
Figure 1. Wavefront Algorithm
judgments based on inaccurate input, which
enables robots to navigate and plan routes in
dynamic and unpredictable settings. Robots can 2. Voronoi Diagrams
now handle ambiguity and vagueness because to
this technique, which makes navigation algorithms Voronoi Diagrams offer a different approach to
more fluid and adaptive. grid-based pathfinding. This method involves
Grid-Based Algorithm - Grid-based algorithms dividing a space into regions or cells, each
are fundamental in path planning and navigation, corresponding to a specific obstacle or point of
particularly in robotics and computer graphics. interest. The edges of these regions represent points
These algorithms discretize the navigation space that are equidistant from the nearest two obstacles,
into a manageable grid, which simplifies the forming what is known as Voronoi edges. These
process of finding paths and navigating around edges constitute paths that maximize the distance
obstacles. This section explores two primary grid- from the nearest obstacles, thereby enhancing
safety in navigation, particularly in scenarios where
collision avoidance is paramount. 1. Dijkstra Algorithm
Developed by Edsger W. Dijkstra in 1956,
Dijkstra's algorithm is a graph search algorithm that
solves the single-source shortest path problem for a
graph with non-negative edge weights. The
algorithm works by iteratively selecting the vertex
with the minimum distance from the source and
updating the distances of its adjacent vertices. It
employs a greedy approach and is often
implemented using priority queues, which can
significantly improve its efficiency. Dijkstra's
algorithm is renowned for its simplicity and
Figure 2. Voronoi Diagrams
effectiveness in many practical applications such as
GPS navigation and network routing.
In path planning, Voronoi Diagrams are
particularly advantageous in open spaces, where
they can effectively navigate around sparse
obstacles while maintaining a safe distance.
However, their utility diminishes in densely
populated areas with numerous obstacles, as the
diagrams become exceedingly complex and
computationally intensive. The computation of
Voronoi Diagrams can be challenging, particularly
in real-time applications or environments with Figure 3. Dijkstra's Algorithm
dynamic changes.
Dijkstra's algorithm begins at the chosen source
Comparing the Wavefront Algorithm and Voronoi node and examines the edges of its adjacent nodes.
Diagrams reveals distinct advantages and In the context of the provided graph, let's choose
limitations of each method. The Wavefront node A as the source. The algorithm initializes the
Algorithm's strength lies in its straightforward distance to the source node as 0 and all other nodes
implementation and reliability in static, structured as infinity. It then proceeds as follows:
environments. It is well-suited for indoor robotics • From node A, the algorithm considers its
navigation and simple game environments. On the
adjacent nodes B and C, with distances 4
other hand, Voronoi Diagrams excel in providing
safer paths in open spaces, making them ideal for and 5, respectively, and updates the
applications like autonomous vehicle navigation in shortest distance to these nodes.
less cluttered environments. • Next, the algorithm selects the node with
Both algorithms demonstrate the versatility and the smallest tentative distance from the
adaptability of grid-based pathfinding methods. source, which is node B in this case, and
They underscore the importance of context in examines its neighbors. It updates the
choosing the most appropriate algorithm for a distance to node D to 13 (4 from A to B
given task. Understanding these algorithms' plus 9 from B to D).
underlying principles, applications, and limitations • The algorithm then moves to the next
enables more informed decision-making in the
closest node, which is C, and updates the
design and implementation of pathfinding and
distance to its neighbor E to 8 (5 from A to
navigation systems.
C plus 3 from C to E).
Graph-Based Algorithm - Graph-based • This process continues, selecting the node
algorithms are fundamental in the field of computer with the smallest known distance at each
science, particularly in solving problems related to step and updating the distances to its
networks and pathfinding. Among these, the neighbors by adding the weight of the
Dijkstra and Bellman-Ford algorithms stand out edge that connects them.
due to their unique approaches to finding the • If a shorter path to a node is discovered,
shortest paths in a graph. This section supplies an the algorithm replaces the previously
overview and comparison of these two algorithms, known distance with the new, shorter
highlighting their applicability and efficiency in distance.
various scenarios.
 • The process repeats until the algorithm has has been found to every node, assuming
found the shortest path to all nodes or the there is no negative weight cycle. For
nodes reachable from the source node. instance, it will update the distance from
node 1 to node 4 through node 2 if the
In the final output, Dijkstra's algorithm provides path 1→2→4 offers a shorter route than
the shortest distance from the source node A to all the current known path.
other nodes in the graph. It also typically stores the • After V−1 iterations, the algorithm will
predecessors for each node, allowing the shortest perform one final check on all edges to
path to be reconstructed from the source node to detect any negative weight cycles. If the
any other node. distance to a node can still be reduced, it
2. Bellman-Ford Algorithm indicates a negative weight cycle exists.
The Bellman-Ford algorithm, on the other hand,
was developed by Richard Bellman and Lester For the given graph, the Bellman-Ford algorithm
Ford independently in 1958. This algorithm is will supply the shortest path from node 1 to all
designed to handle graphs with negative edge other nodes while checking for the feasibility of the
weights, thus offering a broader scope of solution in the presence of negative weight cycles.
application compared to Dijkstra's. It iteratively While both algorithms aim to find the shortest
relaxes edges and updates the cost of the shortest paths in a graph, they differ significantly in their
paths. A significant advantage of the Bellman-Ford operations and applications. Dijkstra's algorithm is
algorithm is its ability to detect negative cycles in a typically faster and more efficient for graphs with
graph, a feature that Dijkstra's algorithm lacks. non-negative weights. In contrast, the Bellman-
However, this comes at the cost of higher time Ford algorithm is more versatile and capable of
complexity. handling graphs with negative edge weights and
detecting negative cycles. The choice between
these algorithms depends largely on the graph's
characteristics and the specific requirements of the
problem at hand.
Sampling-Based Algorithm - In the realm of
robotic motion planning and pathfinding, sampling-
based algorithms have appeared as powerful tools
for navigating complex, high-dimensional spaces.
These algorithms circumvent the need for explicit
Figure 4. Bellman-Ford Algorithm computation of the configuration space by
randomly sampling the space. This section delves
into the nuances of Rapidly-exploring Random
The Bellman-Ford algorithm is adept at handling Trees (RRT), RRT*, and Probabilistic Roadmaps
graphs with negative weight edges and can even (PRM), elucidating their methodologies and
detect negative weight cycles. For the given graph, applications.
assuming node 1 is the source, the algorithm 1. Rapidly-exploring Random Trees (RRT)
proceeds as follows: RRT is a path-planning algorithm renowned for its
efficiency and simplicity. It constructs a space-
• The algorithm initializes the distance to filling tree that incrementally grows from the start
the source node (node 1) as 0 and all other state by randomly sampling points in the search
nodes as infinity. space and extending the tree towards them. The
• It then relaxes all the edges by updating algorithm's strength lies in its ability to quickly
the distance to the destination node if the explore uncharted spaces, making it suitable for
sum of the edge weight and the distance to environments where the path to the goal is not at
the source node is less than the currently once clear. Its probabilistic nature ensures that
known distance. For example, it checks given enough time, it can find a path if one exists,
the edge from node 1 to node 2 (weight 6), albeit without guarantees of optimality.
and node 1 to node 3 (weight 5), updating
the distances accordingly.
• The algorithm continues this process for
each edge, for a total of V−1 iterations,
where V is the number of vertices in the
graph. This ensures that the shortest path
 optimization. As the tree grows, RRT*
continuously refines its branches to approach the
optimal path. This is achieved by considering not
only the nearest neighbor but also a set of nearby
vertices and choosing the parent that will result in
the lowest path cost. Over time, RRT* converges
towards an optimal solution, striking a balance
between exploration and exploitation.
Figure 5. Rapidly-exploring Random Trees (RRT)

The RRT algorithm starts at the 'Start' node and

grows a tree by randomly sampling the space. In
this image, the nodes represent positions in space,
while the lines between them represent feasible
paths that the RRT has explored.

• Initialization: The algorithm initializes the Figure 6. RRT*

tree with the 'Start' node as the root.
• Random Sampling: It then randomly
selects a point in the search space and The four images provided depict the iterative
identifies the nearest node in the tree to process of the RRT* (Rapidly-exploring Random
this random point. Trees Star) algorithm. This algorithm is an
• Node Expansion: From the nearest node, optimized version of RRT and seeks to find the
the algorithm extends the tree towards the most optimal path to the goal as more nodes are
random point by a predefined step size, added. Here is a step-by-step explanation based on
adding a new node to the tree. the images:
• Collision Checking: The paths are • Initialization: The process starts with the
checked for collisions with obstacles, 'Start' node, and the algorithm begins to
represented by the red circles. If a path incrementally build a tree. In the first
segment intersects an obstacle, it is image, we see a few initial nodes
discarded. branching out from the start, showing the
• Growth: This process of random sampling early exploration of space.
and node expansion continues, with the • Random Sampling and Expansion: The
tree rapidly exploring the space and second image shows the tree after more
avoiding obstacles, always growing iterations. The algorithm randomly
towards unexplored areas. samples the space and adds new nodes by
• Reaching the Goal: The algorithm connecting them to the closest vertex in
proceeds until a node is created within a the existing tree, ensuring that the path
predefined distance from the 'Goal' node. does not intersect with any obstacles (red
This final node becomes the goal node's circles).
parent, effectively finding a feasible path • Nearest Neighbor Search and Rewiring: In
from 'Start' to 'Goal'. the third image, the tree has grown
significantly, and the algorithm is now
In the context of the provided image, the blue looking for neighbors within a certain
nodes and green lines illustrate the incremental radius (depicted by the circles) to connect
construction of the tree, with the nodes avoiding the newly added nodes in a way that could
the red obstacle areas and finally reaching the potentially offer a shorter path. The
'Goal' node. RRT does not guarantee the shortest algorithm checks if these new connections
path but is prized for its efficiency and can reduce the path cost to any of the
effectiveness in complex environments. existing nodes and rewires the tree
accordingly.
2. RRT Algorithm* • Path Optimization: By the fourth image,
RRT* builds upon the foundational principles of the tree has reached the 'Goal' node.
RRT with the added capability of path However, the RRT* algorithm doesn’t
 stop at the first connection. It continues to nodes that are within a certain distance of
sample, add, and rewire nodes. During this each other, creating a network of vertices
process, it constantly refines the path, and edges. This process avoids obstacles,
seeking shorter connections and ensuring that paths do not intersect with
optimizing the route from 'Start' to 'Goal' them.
through the iterative process of rewiring. • Roadmap Construction: The result of the
• Convergence to Optimal Path: Over time sampling and connection phases is a
and with sufficient iterations, RRT* roadmap of connected paths that navigate
improves the paths within the tree. It around the obstacles. This roadmap is a
gradually converges towards an optimal graph-like structure that represents
solution that is the shortest path from the possible paths through the space, as seen
start to the goal, given the constraints of in the first image.
the sampled space and the obstacles within • Query: The second image shows the
it. addition of 'Start' and 'End' nodes. When a
The RRT* algorithm is particularly useful in query is made to the PRM for a path from
robotics and autonomous vehicle navigation where start to end, the algorithm connects the
the path planning must be both efficient and as start and end points to the nearest nodes in
optimal as possible given the unpredictable nature the roadmap.
of the environment. The images demonstrate how • Path Finding: Using graph search
the algorithm not only finds a feasible path but algorithms like A* or Dijkstra's, the PRM
refines it toward the optimal path as it progresses. searches the constructed roadmap for the
3. Probabilistic Roadmaps (PRM) shortest path from the start node to the end
PRM is a multi-query path planning algorithm that node. This is represented by the path in
constructs a roadmap of the environment by the second image, which shows a feasible
sampling random configurations and connecting route connecting 'Start' to 'End' while
them through local paths. Once the roadmap is avoiding obstacles.
constructed, the algorithm can answer multiple path • Path Smoothing (optional): The initially
queries by connecting start and goal states to the found path may be suboptimal or jagged,
nearest points on the roadmap and searching for the so path smoothing techniques can be
shortest path. PRM is particularly effective in static applied to create a more direct and
environments and for problems where the efficient route.
computational cost of multiple path queries needs
to be amortized over a single representation of the In summary, the PRM algorithm is a two-phase
space. process where the first phase builds a graph that
captures the connectivity of the free space, and the
second phase queries this graph to find a path
between start and end configurations. This method
is efficient because the roadmap needs to be
computed only once and can be reused for multiple
Figure 7. Probabilistic Roadmaps (PRM) queries, which is beneficial in static environments
The two images illustrate the Probabilistic where the obstacles do not change position.
Roadmap (PRM) method, a popular motion While RRT and RRT* are single-query algorithms,
planning algorithm in robotics and computer-aided excelling in scenarios where a path needs to be
design. PRM is particularly well-suited for high- found quickly without prior knowledge of the
dimensional spaces and complex environments. environment, PRM is more suited for scenarios
Here's how PRM works, as depicted in the images: where the environment is known, and multiple path
• Sampling: The first step in PRM is to queries will be made. RRT* offers a significant
sample random configurations in the advantage over RRT by providing a path that
search space. The first image could asymptotically approaches the optimal solution,
represent this phase where nodes (blue while PRM provides a comprehensive
dots) are placed randomly while avoiding representation of the space that can be reused. The
collision with obstacles (red circles). choice of algorithm thus depends on the specific
• Node Connection: After the nodes are requirements of the application, the nature of the
sampled, the algorithm attempts to environment, and the computational resources
connect these nodes with local paths. The available.
connections (blue lines) are made between Experiments
In robotics, fuzzy logic encompasses various Hybrid Versions of Fuzzy Logic
fundamental ideas:
Fuzzy Sets: These sets permit partial membership,
in contrast to conventional sets that have binary
membership (an element is either in or out). This is
an example of ambiguous or uncertain data, which
is necessary for practical uses.

Degree of Membership: In a fuzzy set, the degree

of membership of each element ranges from 0 to 1.
This helps robotics decision-making to be more
nuanced by indicating the degree of truth or the
extent to which an element is part of a set. Figure 8. Collaboration of different types of algorithms
with fuzzy logic
Fuzzy Rules and Inference: Conditional statements
that define relationships using linguistic variables
are known as fuzzy rules. Because a fuzzy Hybridization with neural networks -
inference system processes these principles, robots The goal of robotics' hybridization of fuzzy logic
can infer actions from imprecise inputs. and neural networks is to improve robots' path-
Defuzzification: The last stage of fuzzy logic planning abilities. This combination improves
transforms fuzzy set outputs into a single, sharp decision-making in complicated contexts by
value, giving the robot a clear course of action or combining the interpretability of fuzzy logic with
decision. Practical applicability in real-world the adaptive learning capabilities of neural
situations is ensured by this method. networks. These systems are appropriate for real-
time applications in both static and dynamic
Working environments since they can learn more quickly
and require less time and money to create. With
Static environment - Robots can travel in static increased computational efficiency, they navigate,
settings with the use of fuzzy logic, which optimize pathways, and adjust to changing
interprets layout limitations and fixed barriers. conditions by using fuzzy rules that are based on
Since static environments are immutable, as neural learning.
opposed to dynamic ones, the fuzzy logic system
concentrates on path optimization using Hybridization with Multi-Agent Systems - Fuzzy
predetermined parameters. The robot utilizes fuzzy logic hybridization with multi-agent systems
rules to make judgments, taking into account the (MAS) leverages agent collective behavior to
level of safety and proximity to barriers. Fuzzy sets improve robotic path planning. Every joint in a
are used to represent environmental components robot is thought of as an agent with certain
such as walls or stationary gear. This enables the objectives, including lining up with a target or
robot to select the most effective, collision-free dodging obstacles. Each agent behaves
course even in complex environments or situations independently yet cooperatively through the use of
when accurate data are not available. fuzzy logic controllers (FLC), allowing a global
behavior to evolve for efficient path planning.
Dynamic environment - Fuzzy logic is essential Experiments with robotic manipulators navigating
for real-time adaptability in changing contexts. around obstacles show that this approach is ideal
Robots may continually reevaluate and modify for real-time applications and enables for dynamic
their routes thanks to fuzzy logic, even in the adjustment to unknown and new surroundings.
presence of changing situations or moving objects.
The system uses fuzzy rules to interpret changes in Hybridization with Particle Swarm
the dynamics of the environment by processing Optimization - The interpretive power of fuzzy
sensor data. This leads to an adaptable decision- logic and the powerful searching capabilities of
making process, which helps robots navigate safely particle swarm optimization (PSO) are combined in
in unpredictable environments. Continually the hybridization of the two techniques. PSO
updated to reflect real-time changes, the degree of assists in optimizing fuzzy logic controller
membership in fuzzy sets facilitates informed and parameters, particularly the antecedent and
flexible path-planning decisions. consequent parameters of fuzzy logic systems of
the Mamdani type. Avoiding local minima, a
common optimization issue is made especially online methods combine obstacle avoidance and
successful by this synergy. The hybrid approach fuzzy path planning to allow robots to continue
has proven to be more reliable and efficient for operating and pursuing goals despite joint or
trajectory control in robotic route planning than mechanical problems. These techniques—validated
conventional methods by utilizing cost functions by simulations—are essential for reliable end-
such as Mean of Root of Squared Error and others. effector performance across a range of models.

Hybridization with probabilistic roadmap Part Assembly - Robotics part assembly tasks have
methods - Fuzzy systems are combined with led to the development of fuzzy logic techniques
PRM's stochastic path planning in the hybridization that concentrate on path planning in environments
of fuzzy logic with probabilistic roadmap methods with both static and dynamic impediments. Fuzzy
(FL-PRM) to address uncertainty in robotics minimal entropy, which evaluates variance and
navigation. In Nielsen and Kavraki's method, uncertainty during assembly, is one method that has
complicated path planning is broken down into improved decision-making under uncertainty. Other
more manageable point-to-point problems using a methods combine fuzzy logic and neural networks
fuzzy roadmap. These problems are then resolved (FL-NN). These techniques, which are supported
using fuzzy PRM planners. Experiments with by simulations and real-world trials using cutting-
robotic manipulators handling huge objects verify edge sensors, show how to execute tasks efficiently
this two-level system, which reduces collision in path planning, increasing the precision and
checks and boosts efficiency. By using this method, dependability of robotic manipulators throughout
robots in static situations can be more adaptable assembly processes.
and have more reliable path planning.
Pick and place/object grasping tasks - Hybrids of
Hybridization with rapidly-exploring random neural networks (NN) and fuzzy logic (FL)
trees - Fuzzy logic and the quickly developing increase the efficiency of manipulator robots in
random tree (RRT) technique are hybridized when pick-and-place and object-grasping tasks. To ensure
fuzzy systems are used to improve the RRT's accurate movements, these systems employ fuzzy
performance in static settings. Fuzzy logic, which logic to correct for flaws in inverse kinematics
is more adept than classical RRT in determining the calculations. By using sensors and making real-
closeness to barriers and probable collisions, can be time movement adjustments, they improve robots'
used to evaluate cost functions in a more nuanced capacity to recognize things and carry out
manner because to this fusion. The modified activities, such as picking tomatoes, demonstrating
method, called Transition-based RRT (T-RRT), the efficacy of FL in terms of robotic precision and
provides manipulator robots with a more advanced stability.
navigation solution in environments with stationary
obstacles by making the path-planning process Experimental results in papers
more flexible to the geometrical complexities of the
surroundings. Paper 1
Problem statement - The high dimensionality
Applications configuration space (C-space) challenge in
collision-free path planning for manipulator robots
Path Optimization - Path optimization is essential (CFPPMR) is the paper's stated problem. It entails
for achieving maximum energy and time economy developing a path planner algorithm that, while
in robotic manipulator systems. Fuzzy logic helps avoiding static and dynamic impediments, abiding
improve motion efficiency and kinematic model by mechanical limitations, and taking into account
refinement; nonetheless, rule quality plays a major potential joint failures, connects a robot's starting
role in path optimality. Hybrid approaches, such as location and orientation (Source) to a desired
fuzzy logic combined with evolutionary configuration (Target).
algorithms, have the potential to optimize energy
usage while accounting for stochastic parameters Results - The experimental findings demonstrate a
such as friction, especially in robots with limited mobile robot with ultrasonic sensors that uses a
degrees of freedom. minimum-risk strategy to navigate through
uncharted territory. The robot varies the impact of
Fault Tolerance - Fuzzy logic techniques are used goal-seeking (GS) and obstacle avoidance (OA)
in robotic fault tolerance to guarantee operational behaviors by adjusting its speed and trajectory in
continuity in the event of mechanical faults. These reaction to obstacles. One can exhibit effective
 Figure 9. Simulation result for fuzzy logic integrated with path searching algorithm

navigation in intricate settings, such as spaces with

shifting obstructions and U-shaped routes. Even in Fuzzy logic is used by a mobile robot to navigate
the presence of possible sensor drift, the method by analyzing sensor data and modifying its motor
works well, underscoring the significance of speeds to avoid obstacles. The system's objective is
adaptive behavior and risk mitigation in robotic to efficiently reach a target by adjusting direction
path planning. and speed in response to the proximity of obstacles.
With 27 rules, the fuzzy controller guarantees
Paper 2 motion free from collisions, as proven by
Problem statement - The research takes into simulations involving a range of obstacle
account the high dimensionality of the situations.
configuration space (C-space) when building path
planners for the Collision-Free Path Planning for Paper 3
Manipulator Robots (CFPPMR). Finding a solution
that utilizes data on the explicit geometric Problem statement - The difficulty of creating
representation of the free C-space and ensures the route planners for Collision-Free route Planning for
translation and rotation of the robot's end-effector Manipulator Robots (CFPPMR) because of the
from the starting position (Source) to the desired configuration space's high dimensionality (C-
configuration (Target), while staying within space) is the paper's stated concern. Its main
mechanical constraints and dodging obstacles, is objective is to create a system that links a robot's
frequently made more difficult by this complexity. initial configuration (Source) to its intended
configuration (Target), taking into account
Results mechanical limitations and the requirement to
avoid both static and dynamic impediments.

Results

Figure 10. Simulation result (varied obstacle size) Figure 11. simulation result
The outcomes of the experiment verified that a Mobility-Assisted Localization in Wireless Sensor
mobile robot with ultrasonic sonars may Networks. Sensors 2017, 17,1904.
successfully navigate across an unknown https://doi.org/10.3390/s17081904
environment with local minimums when using the
[6] Meng Wang and Liu, "Fuzzy logic based robot
minimum risk strategy. The robot was able to avoid
path planning in unknown environment," 2005
collisions and escape from local minimums by International Conference on Machine Learning and
detecting obstructions within three meters of its Cybernetics, Guangzhou, China, 2005, pp. 813-818
path and modifying its route accordingly. Vol. 2, doi: 10.1109/ICMLC.2005.1527055.

Paper 4 [7] Horacio Martı́nez-Alfaro, Simón Gómez-

Problem statement - The paper's problem Garcı́a,Mobile robot path planning and tracking
statement centers on developing a path planner to using simulated annealing and fuzzy logic
guide a robot from its initial configuration to its control,Expert Systems with Applications,Volume
intended destination. It is challenging to identify 15, Issues 3–4,1998,Pages 421-429,ISSN 0957-
solutions based on explicit geometric 4174,https://doi.org/10.1016/S0957-
representations of the free C-space while taking 4174(98)00055-4.
mechanical restrictions and avoiding obstructions
[8] B. Li and B. Chen, "An Adaptive Rapidly-
into account due to the complexity of the high-
Exploring Random Tree," in IEEE/CAA Journal of
dimensional configuration space (C-space). Automatica Sinica, vol. 9, no. 2, pp. 283-294,
February 2022, doi: 10.1109/JAS.2021.1004252.
Results - The main objective of the experimental
results was to assess the accuracy, precision, and [9] A. Pandey, R. K. Sonkar, K. K. Pandey and D.
localization ratio of the suggested model to R. Parhi, "Path planning navigation of mobile robot
determine its efficiency. The average localization with obstacles avoidance using fuzzy logic
error and its standard deviation among models controller," 2014 IEEE 8th International
served as a proxy for localization accuracy, a Conference on Intelligent Systems and Control
crucial component of path planning. Reducing (ISCO), Coimbatore, India, 2014, pp. 39-41, doi:
localization error was the aim to increase the 10.1109/ISCO.2014.7103914.
model's accuracy.
[10] Hentout, A., Maoudj, A. & Aouache, M. A
References review of the literature on fuzzy-logic approaches
for collision-free path planning of manipulator
[1] Roadmap-Based Path Planning - Using the robots. Artif Intell Rev 56, 3369–3444 (2023).
Voronoi Diagram for a Clearance-Based Shortest https://doi.org/10.1007/s10462-022-10257-7
Path DOI: 10.1109/MRA.2008.921540
[11] Minguez, Javier and Luis Montano. “Nearness
[2] Voronoi Diagrams — A Survey of a diagram (ND) navigation: collision avoidance in
Fundamental Geometric Data Structure, FRANZ troublesome scenarios.” IEEE Transactions on
AURENHAMMER DOI: 116873.116880 Robotics and Automation 20 (2004): 45-59.

[3] Parimala, M., Broumi, S., Prakash, K. et al. [12] Javaid, Muhammad Adeel, Understanding
Bellman–Ford algorithm for solving shortest path Dijkstra's Algorithm (April 10, 2013). Available at
problem of a network under a picture fuzzy SSRN: https://ssrn.com/abstract=2340905 or
environment. Complex Intell. Syst. 7, 2373–2381 http://dx.doi.org/10.2139/ssrn.2340905
(2021). https://doi.org/10.1007/s40747-021-00430-
w

[4] Rodriguez, Xinyu Tang, Jyh-Ming Lien and N.

M. Amato, "An obstacle-based rapidly-exploring
random tree," Proceedings 2006 IEEE International
Conference on Robotics and Automation, 2006.
ICRA 2006., Orlando, FL, USA, 2006, pp. 895-
900, doi: 10.1109/ROBOT.2006.1641823.

[5] Alomari, A.; Phillips, W.; Aslam, N.; Comeau, F.

Dynamic Fuzzy-Logic Based Path Planning for