Page 254 - 2024-Vol20-Issue2

250 |
algorithm uses the Kalman filter. This filter uses sensor data
to accurately estimate the robot’s position, reducing errors.
Smoothing the path allows the GA-PRM find a balance be-
tween avoiding collisions and smooth movement. The Kalman
filter updates the robot’s position based on sensor data, en-
suring the robot can navigate safely and efficiently, even in
changing environments, without putting safety at risk or dis-
rupting medical tasks.
The GA-PRM algorithm, as described in the flowchart
in Figure 2, follows a step-by-step process for efficient path
planning. It begins with initialization and then utilizes the
Probabilistic Roadmap (PRM) technique to create a roadmap
by randomly sampling and connecting configurations. Subse-
quently, it generates a path segment from the current position
towards the goal, considering forces to navigate obstacles.
The generated path’s feasibility is checked for goal reachabil-
ity. To optimize the path, a Genetic Algorithm is employed,
creating and refining potential paths within a population. Con-
tinuous goal checking occurs throughout execution. Dynamic
obstacles (people) are introduced by randomly moving them
within the workspace to simulate dynamic environments. The
Kalman Filter enhances path estimation accuracy and mini-
mizes sensor measurement noise. The algorithm concludes its
execution, providing efficient path planning while accommo-
dating obstacles and dynamic scenarios.
D. Modifications and Improvements
In this section, the modifications and improvements made to
existing path-planning algorithms to develop a novel approach
are described. These enhancements aim to address specific
limitations and leverage the strengths of the individual algo-
rithms. The modifications are designed to work together and
perfectly within the framework of the proposed algorithm.
1. Combining Strengths: The algorithm blends Genetic
Algorithm (GA) and Probabilistic Roadmaps (PRM) in
a unique way. GA explores possibilities, while PRM
carefully maps valid paths. They work together: GA
suggests paths, and PRM refines them. This combina-
tion improves path quality and speeds up finding the
best path.
2. Faster Computing: Smart data handling, like the R-
tree index, speeds up PRM’s work, making it suitable
for quick path planning. Additionally, advanced meth-
ods like the Kalman filter improve sensor data process-
ing, reducing the workload and making it even better
for real-time tasks.
3. Balancing Goals: The algorithm considers various fac-
tors like path length, smoothness, and computation time
all at once. This flexible approach gives different op-
tions, allowing users to balance these factors as needed.
Sabeeh & Al-Furati
Using an R-tree data structure [39] significantly boosts essen-
tial tasks like finding nearby positions and checking for col-
lisions. This unique data structure, the R-tree index, smartly
organizes spatial data. It does this by creating an efficient sys-
tem where information is stored in a way that allows extremely
fast access. This dynamic indexing method is a significant
advancement, especially for real-time applications. Unlike
older indexing methods, which struggle with objects of vary-
ing sizes in complex, multi-dimensional spaces, the R-tree
brings a simplicity that greatly enhances the robot’s ability to
navigate swiftly and accurately.
IV. PROBLEM FORMULATION
This propose approach aims to find the best path for a mo-
bile robot in a workspace with both stationary and moving
obstacles. This involves finding a sequence of grid cells that
connect the start and end points while avoiding collisions with
obstacles. This section formally defines the path-planning
problem and describes the system model.
A. Path-Planning Formalization
Path-planning is the process of finding a safe and efficient
path for a robot to move from one point to another in an
environment. The environment may contain obstacles, such
as walls and furniture, as well as dynamic obstacles, such
as people and vehicles. The path-planning problem can be
broken down into the following steps:
• Representation of the workspace: The workspace is
represented as a grid-based environment. This means
that the environment is divided into a grid of cells, and
each cell represents a specific location in the workspace.
This representation makes it easier for the robot to plan
its path and avoid obstacles.
• Definition of the start and goal points: The start
point is the robot’s initial position, and the goal point
is the desired destination. The path-planning algorithm
must find a path that connects the start and goal points
without colliding with any obstacles.
• Identification of static obstacles: Static obstacles are
obstacles that do not move during the planning process.
Their coordinates on the grid identify these obstacles,
and they are marked as such in the grid representation.
• Accounting for dynamic obstacles: Dynamic obsta-
cles are obstacles that move during the planning process.
These obstacles are represented by their respective mo-
tion models, which describe how they move over time.
The path-planning algorithm must take into account the
motion models of dynamic obstacles when planning the
robot’s path.