EEE F411: Internet of Things

Comprehensive Report

Group 3

Problem Statement: Cooperative Detection and Communication of Static

Obstacles for Enhanced UAV Swarm Navigation

Members:
- Abhinav Reddy Kallem : 2021B4A32408H
- Abhiram Juturu : 2021B4A83143H
- Anmol Murti : 2021B4A82539H
- CH V V Sai Prajeet : 2021AAPS3014H
- Rohan Reddy Amanaganti : 2021B4A32377H
- Venkata Teja Pappu : 2021B4A82968H

1
 CONTENTS
CONTENTS 2
1 - Problem Statement 4
1.1 - Statement 4
1.2 - Introduction 4
2 - Methodology and Framework 5
2.1 - Obstacle Detection 5
2.1.1 - Sensor Selection 5
2.1.2 - Sensor data fusion 5
2.1.3 - Object Detection Algorithm 6
2.2 - Communication Protocols 6
2.2.1 - Data Sharing Strategy 6
2.2.2 - Communication Methods 7
2.3 - Path Planning 7
2.3.1 - Decentralized Path Planning 7
2.3.2 - Cooperative Path Planning 7
3 - Solution and Results 8
3.1 - Proposed solution 8
3.1.1 - UAV Communication and Obstacle Sharing 8
Detection and Sharing 8
Communication Protocol 9
3.1.2 - Periodic Synchronization 9
Mechanism 9
Benefits 9
3.1.3 - Relay Communication 10
Extended Coverage 10
Dynamic Relaying 10
3.1.4 - Centralized vs. Distributed Communication 10
Distributed Communication 10
Centralized Communication 10
3.1.5 - Energy Consumption Modeling 11
Message-Based Energy Model 11
Optimization Strategies 11
3.1.6 - Performance Metrics 11
Latency 11
Obstacle Sharing Efficiency 11
Detection Accuracy 11
Energy Consumption 12
Scalability Metric 12
3.1.7 - Advantages of the Proposed Solution 12

2
 3.2 - Simulation Approach Overview 12
3.2.1 - Hybrid Communication (Proposed Solution) 13
Mechanics 13
Advantages 13
Disadvantages 13
Ideal Use Case 13
3.2.2 - Decentralized Communication with Single Sensor 13
Mechanics 13
Advantages 14
Disadvantages 14
Ideal Use Case 14
3.2.3 - Centralized Communication 14
Mechanics 14
Advantages 14
Disadvantages 14
Ideal Use Case 15
3.2.4 - Distributed Communication with Kalman Filtering 15
Mechanics 15
Advantages 15
Disadvantages 15
Ideal Use Case 15
3.2.5 - Comparative Overview 16
3.3 - Simulation Configuration 16
3.3.1 - Environment 17
3.3.2 - Experimental Constants 17
3.3.3 - Experimental Parameters (For Comparision Basis) 17
3.4 - Simulation Results 18
3.4.1 - Hybrid Communication (Proposed Solution) 18
3.4.2 - Decentralized Communication with Single Sensor 18
3.4.3 - Centralized Communication 18
3.4.4 - Distributed Communication with Kalman Filtering 18
3.4.5 - Environment Plots of Randomly Generated Objects during Simulation Process 19
4 - Conclusion 21
5 - References 22
6 - Group Members and Work Distribution 23

3
1 - Problem Statement

1.1 - Statement
Cooperative Detection and Communication of Obstacles for Enhanced UAV Swarm Navigation

1.2 - Introduction
As the use of UAV swarms across various industries keeps rising, the need for reliable and
efficient obstacle detection becomes increasingly crucial, especially in extremely harsh
environments rich in obstacles and highly unpredictable.

Traditionally, UAVs rely on onboard sensor data to identify surrounding threats and modify flight
plans accordingly. However, this can become unreliable with swarms, where high levels of
coordination are necessary for effective operation. Individual detection becomes highly risky,
considering delays in processing and relying on all the drones to identify all surrounding
obstacles at all times, which cannot be feasible in cases where drones may be outside the
obstacle detection range but are on track to collide with one that had been detected by another
potentially or if a UAV has compromised sensors.

To address this, we have proposed the idea of cooperative detection and communication of
static obstacles among UAVs in a swarm. The fundamental idea behind this is to have a UAV
that has detected the accurate location and nature of a static obstacle and relay this information
to other UAVs around it within the same swarm. By doing so, the neighboring UAVs can plan an
alternate route in real time. This allows for efficient and coordinated swarm operation in dense
and harsh environments.

Therefore, the core objectives of our proposed system are

- Reliable Detection: We must develop a robust system for individual UAVs that can detect
static objects, no matter what environmental conditions they are present in (rain, fog,
poor lighting, etc.). It, therefore, requires the development of sensor integration, sensor
fusion, and obstacle detection algorithms and must be present onboard each UAV and
operating in real-time.

- Effective communication and Processing: We must develop a low-latency

communication protocol to enable UAVs within a swarm to transfer obstacle data with
one another. This system must be scalable, operate efficiently at long ranges without
transmission losses, and resist interference to ensure all UAVs are up to date with the
obstacles around them, can keep track of these, and act accordingly.

In this report, we will go over potential approaches to the objectives mentioned above, compare
these with one another, and simulate our proposed solution to one of these subsystems

4
2 - Methodology and Framework
Here, we can break down this approach to solving the problem statement into three
subproblems.

- Obstacle Detection
- Communication Protocols
- Path Planning and Decision Making

2.1 - Obstacle Detection

In this subsystem, we rely on each UAV to identify static obstacles with high levels of accuracy,
given any scenario. This subsystem consists of 3 components for effective functioning.

2.1.1 - Sensor Selection

Sensor selection can become crucial in object detection as each has strengths and drawbacks
in certain environments. These include

- Ultrasonic Sensors: These work by emitting and capturing sound waves to calculate the
distance between the UAV and any nearby obstacle. This works effectively in
short-range detection scenarios to find the proximity of any obstacle in a scenario where
a UAV may be traveling slowly through tight spaces
- Optical Cameras and Stereo Vision: These provide visual data for computer vision (CV)
algorithms to use and process and pinpoint if any obstacle is detected. Stereo vision
uses multiple cameras in tandem to allow depth perception and map these objects in a
three-dimensional space. This works well in almost all environments where vision isn’t
hindered
- LiDAR or RGB-D sensors: These help the drones estimate the height and distance of
obstacles more precisely through laser beams for measuring distances. This, therefore,
allows one to create a 3D map of the surroundings and further improves their ability to
navigate safely in three-dimensional space.
- Infrared Cameras: These work by detecting infrared emission around the surroundings,
even in low visibility areas (fog, at night, etc.)

2.1.2 - Sensor data fusion

Since each sensor has its unique set of advantages and disadvantages, we can use multiple
types of sensors in tandem and combine the data obtained from each to create a
comprehensive map of the surroundings through sensor fusion, therefore allowing higher
accuracy detection.

- Kalman Filters: This merges obtained sensor data to create a unified estimate of
obstacle coordinates while simultaneously accounting for any sensor data discrepancies.

5
 - Particle Filters and SLAM (Simultaneous Localization and Mapping): This continuously
captures its surroundings to create a map and positions the UAV’s current location onto
this. This is useful in unknown territories, where the UAV must look for new obstacles.

2.1.3 - Object Detection Algorithm

YOLOV5 i.c.w. SOFT-NMS:

This is a popular object detection algorithm with high speed and accuracy with the ability to
detect multiple obstacles at a time. It works based on using a single forward pass of a neural
network to predict both the class and bounding boxes of the objects present in an image

The enhanced YOLOv5 neural network with an improved Non-Maximum Suppression algorithm
(SOFT-NMS) uses this post-processing tool to filter out overlapping bounding boxes by only
keeping those with a high confidence score, allowing better detection. The improvements made
YOLOv5 scalable, lightweight, accurate, and suitable for detecting static obstacles.

This consists of 2 key components

- Gaussian Filtering: Preprocessing of images to minimize noise.

- Attention Mechanism: Added to improve the focus on important image areas for better
detection accuracy.

Advantages

- Enables real-time detection.

- Suitable for low-elevation and urban environments.

- Open-source, easy to upgrade.

2.2 - Communication Protocols

2.2.1 - Data Sharing Strategy

- Centralized Processing: Here, one drone acts as the chief and is equipped with the
processing capabilities of YOLOv5. The rest of the swarm sends information to other
UAVs vis the chief, freeing up processing power on the remaining drones and reducing
communication load.

- Distributed Processing: Here, each drone in the swarm has the capability to run
YOLOv5, and each share its findings through a low-latency communication protocol.
This allows each drone to process its surroundings in real-time and make premeditated
decisions for avoidance

6
2.2.2 - Communication Methods
- LoRa (Long Range Communication): For long-distance communication, LoRa can
transmit critical information, such as detected obstacles, between drones that are far
apart. This helps with tasks like long-range navigation.

- Bluetooth Mesh or Wi-Fi: For short-range communication, where drones are closely
flying together, Bluetooth Mesh or Wi-Fi can be used. This allows drones to share data
about nearby obstacles and coordinate better without the need for a centralized
infrastructure.

- Cellular (5G/6G): This allows for long-range transmission over cellular networks, often
with high bandwidth and low latency, similar to mobile networks. Due to the already
established infrastructure, this makes cellular a highly viable method of communication

2.3 - Path Planning

Based on the obstacles detected by onboard sensors or neighboring UAVs, each UAV must
reroute itself onto a more optimal path to avoid imminent dangers. This requires path-planning
algorithms to handle the rerouting of UAVs to traverse with minimal hindrances. These can be
divided into 2 categories.

2.3.1 - Decentralized Path Planning

Here, each UAV can make decisions and act independently based on the obtained obstacle
data.

- A* Algorithm: This popular graph-based algorithm works based on finding the shortest
possible path to the destination while avoiding obstacles. Each UAV can run this
algorithm to find its unique shortest path to reach the destination.

- D* Algorithm: This is an extension to the A* algorithm, which accounts for the map
changes over time as the UAV keeps feeding in new data it detects.

2.3.2 - Cooperative Path Planning

Here, the UAVs process individually their preliminary trajectories and then broadcast these to
the others in the swarm to avoid any conflicts if present
- Distributed Path Planning: Here, UAVs share their flight plans, pinpoint if any overlaps
exist, and correct these to ensure that no 2 drones end up colliding with one another or
end up in too close vicinity.
- Swarm Intelligence Algorithms: Here, algorithms like PSO (Particle Swarm Optimization)
allow for collaborative decision-making in path planning, where the UAVs collectively
reach a consensus on an optimal plan that ensures a steady formation is maintained.

7
3 - Solution and Results
Below, we discuss our solution to at least one of the above-mentioned subsystems and simulate
and compare existing state-of-the-art solutions.

3.1 - Proposed solution

Hybrid Communication for UAV Swarm Obstacle Detection

The hybrid communication protocol is designed to address the challenges of cooperative
obstacle detection and sharing in UAV swarms operating in complex environments. It combines
the strengths of distributed communication, relay-based data sharing, and periodic
synchronization to ensure reliable and scalable operation. The system integrates advanced
features such as sensor fusion and low-latency communication to enhance detection accuracy
and obstacle propagation efficiency.

This solution overcomes the limitations of purely decentralized or centralized systems by

introducing a cooperative framework that balances latency, energy efficiency, and reliability.

Discussed below are all the key features involving this solution with comparisions

3.1.1 - UAV Communication and Obstacle Sharing

Detection and Sharing

- Each UAV is capable of detecting obstacles in its vicinity due to the presence of onboard
sensors. In our simulation, we have taken the detection probability factor of 0.5 to
account for faulty sensors or inaccurate capture of obstacles due to environmental
conditions.
- Upon deteting an obstacle, each UAV logs this in a register with the following details
- Obstacle ID: A unique identifier for the obstacle
- Coordinates: 3D location of the obstacle
- Detection Confidence: A confidence score which can be enhanced by sensor
fusion
- Upon detection, these obstacles are then shared to nearby UAVs within a defined
communication radius (for simulation purposes, we have taken this as 40 meters,
although this depends on communication models used in real life)
- This communication radius allows UAVs to exchange data only when in close proximity
with each other, ensuring that all the UAVs are able to map their surroundings

8
Communication Protocol
- The communication between UAVs is designed to minimize latency and ensure reliable
data transfer.
- Each UAV maintains a log of known obstacles, distinguishing between direct detections
and shared obstacles received from other UAVs.
- Shared data includes both obstacle information and the ID of the detecting UAV,
enabling traceability.

3.1.2 - Periodic Synchronization

Mechanism
- Periodic synchronization ensures that all UAVs maintain a consistent and up-to-date
view of the environment.
- UAVs synchronize their obstacle data with all other UAVs within range. This mechanism
reduces discrepancies in obstacle information caused by delays or missed
transmissions.
- Synchronization is particularly effective in scenarios where UAVs temporarily lose direct
communication due to movement or environmental interference.

Benefits
- Minimized Latency: By periodically syncing data, the swarm reduces the time required
for all UAVs to become aware of new obstacles.
- Enhanced Redundancy: If a UAV fails to receive data during one step, it can obtain the
missing information during synchronization.

9
3.1.3 - Relay Communication

Extended Coverage
- UAVs with a larger communication range (or strategically positioned UAVs) act as relays,
forwarding obstacle data to UAVs beyond the range of direct communication.
- Relay UAVs bridge gaps in the swarm’s communication network, ensuring that all UAVs
remain informed even in large or dispersed formations.

Dynamic Relaying

- Relay roles are dynamically assigned based on proximity and communication

capabilities, reducing energy consumption by preventing redundant transmissions.

3.1.4 - Centralized vs. Distributed Communication

The system supports two distinct modes of communication:

Distributed Communication

- Each UAV independently shares obstacle data with nearby UAVs.

- This approach promotes robustness and scalability, as it eliminates dependency on a
single UAV.
- Suitable for dynamic or unpredictable environments where UAV positions change
frequently.

Centralized Communication

- One UAV is designated as the chief UAV, responsible for collecting obstacle data from all
other UAVs.

10
 - The chief UAV consolidates the data and disseminates it back to the swarm.
- While this approach ensures complete data synchronization, it is more susceptible to
failures if the chief UAV is compromised.

3.1.5 - Energy Consumption Modeling

Message-Based Energy Model

- The system models energy consumption based on the number of messages exchanged
during communication.
- Each transmission incurs a fixed energy cost (e.g., 0.1 units/message). The total energy
consumption is computed by summing the energy costs across all UAVs over the
simulation duration.

Optimization Strategies

- Relay Efficiency: By minimizing redundant transmissions, relay mechanisms help

conserve energy.
- Selective Sharing: UAVs prioritize sharing new or high-confidence obstacle data,
reducing unnecessary communication.

3.1.6 - Performance Metrics

The following metrics are tracked to evaluate the system's performance:

Latency

- Measures the time taken for obstacle information to propagate through the swarm.
- Reduced latency is achieved via periodic synchronization and efficient relaying.

Obstacle Sharing Efficiency

- Calculated as the ratio of successfully shared obstacles to the total number of detected
obstacles.
- A high efficiency (>99%) indicates that most obstacles detected by one UAV are quickly
shared with others.

Detection Accuracy

- Compares the number of detected obstacles to the expected total based on the
detection probability and number of UAVs.
- Accuracy values close to 100% indicate the system's reliability in identifying obstacles.

11
Energy Consumption

- Represents the total energy expended in communications. Lower values indicate better
energy efficiency.

Scalability Metric

- Evaluates the system’s ability to handle increasing numbers of UAVs and simulation
steps without significant degradation in performance.

3.1.7 - Advantages of the Proposed Solution

- Real-Time Collaboration: The combination of detection, sharing, and synchronization

allows UAVs to collaborate effectively in real time.
- Redundancy: Periodic synchronization and relaying mechanisms ensure resilience
against communication failures.
- Scalability: The distributed communication mode supports large UAV swarms without
requiring centralized coordination.
- Energy Efficiency: Optimized communication protocols reduce unnecessary energy
usage, prolonging UAV operational time.

3.2 - Simulation Approach Overview

For this solution, we have developed a simulation of our proposed solution (Hybrid
communication) and put it up against 3 state of the art approaches.

12
3.2.1 - Hybrid Communication (Proposed Solution)

Mechanics

The hybrid communication approach combines elements of distributed communication, relaying,

periodic synchronization, and sensor fusion to create a robust and scalable system. Each UAV
independently detects obstacles using onboard sensors and shares the data with its neighbors.
A relay mechanism extends communication beyond direct neighbors, while periodic
synchronization ensures data consistency across the swarm. Advanced sensor fusion
algorithms, such as Kalman filters, enhance detection accuracy by combining inputs from
multiple sensors.

Advantages

- Wide Coverage: The relay mechanism ensures that obstacle data reaches UAVs outside
the direct communication range.
- High Accuracy: Sensor fusion reduces false positives and improves confidence in
detections.
- Data Consistency: Periodic synchronization resolves discrepancies and ensures the
entire swarm has up-to-date obstacle information.
- Scalability: Efficiently supports larger swarms through distributed processing.

Disadvantages

- Moderate Latency: Relay and synchronization mechanisms introduce slight delays

compared to purely decentralized systems.
- Energy Overhead: Additional communication steps (relay and synchronization) consume
more energy than simpler systems.

Ideal Use Case

The hybrid communication approach is ideal for UAV swarms operating in dense, obstacle-rich
environments where wide coverage, accuracy, and consistency are crucial. It excels in
scenarios requiring high coordination and reliable obstacle data propagation.

3.2.2 - Decentralized Communication with Single Sensor

Mechanics

In decentralized communication, each UAV operates independently without sharing obstacle

data with other UAVs. Each UAV relies solely on its onboard sensors to detect obstacles and
make navigation decisions. There is no data sharing, relaying, or centralized processing.

13
Advantages

- Low Latency: No communication delays since UAVs operate independently.

- Energy Efficiency: Minimal energy consumption as no inter-UAV communication occurs.
- Scalability: Easily scales to larger swarms since each UAV functions autonomously.

Disadvantages

- Limited Data Availability: UAVs only know obstacles within their sensing range, leading
to gaps in the swarm's overall situational awareness.
- Increased Risk: Obstacles detected by one UAV are not communicated to others,
increasing the likelihood of collisions or navigation errors in dynamic environments.

Ideal Use Case

This approach works best in open environments with minimal obstacles where UAVs can
operate independently without requiring extensive coordination.

3.2.3 - Centralized Communication

Mechanics

In centralized communication, a single UAV (designated as the chief UAV) collects obstacle data
from all other UAVs, processes it, and redistributes the consolidated data back to the swarm.
The chief UAV is the sole decision-maker responsible for maintaining a unified obstacle
database.

Advantages

- Unified Data Processing: Centralized aggregation ensures all UAVs can access the
same obstacle data.
- Simplified UAV Operations: Non-chief UAVs only need to detect obstacles and send data
to the chief, reducing their computational load.
- Structured Coordination: Centralized processing allows for streamlined decision-making
and navigation planning.

Disadvantages

- High Latency: The need to aggregate and redistribute data creates significant delays,
especially in large swarms.
- Single Point of Failure: If the chief UAV fails or is compromised, the entire swarm's
communication system collapses.
- Energy Consumption: The chief UAV consumes significantly more energy due to its
central role, limiting its operational lifespan.

14
Ideal Use Case

Centralized communication is suitable for small swarms operating in controlled environments

where communication delays and potential failures are less critical. It is also useful when UAVs
have limited computational capabilities and rely on the chief for decision-making.

3.2.4 - Distributed Communication with Kalman Filtering

Mechanics

Distributed communication with Kalman filtering involves all UAVs independently detecting
obstacles and sharing this data with their neighbors. The shared data is then processed using
Kalman filters to create an accurate and unified obstacle map for each UAV. Kalman filtering
combines sensor inputs while accounting for noise and uncertainties.

Advantages

- High Accuracy: Kalman filtering improves detection reliability by merging data from
multiple sources.
- Decentralized Coordination: No reliance on a single UAV, ensuring the system is robust
to individual failures.
- Low Latency: Real-time data sharing and fusion minimize delays.

Disadvantages

- Moderate Energy Consumption: Continuous communication and processing require

more energy than decentralized systems.
- Inconsistent Data: Without periodic synchronization, discrepancies may arise between
the obstacle databases of different UAVs.
- Computational Overhead: Each UAV must run Kalman filters, increasing onboard
processing demands.

Ideal Use Case

This approach is ideal for medium-sized swarms operating in environments with critical obstacle
detection accuracy, but the communication range is sufficient to maintain direct links between
UAVs.

15
3.2.5 - Comparative Overview

Feature Hybrid Decentralized Centralized Distributed w/

Communication Communication Communication Kalman
Filtering

Data Sharing Full (relay & None Via chief UAV Neighbor-to-ne
sync) ighbor

Latency Moderate Low High Moderate

Accuracy High Low High High

Energy Moderate Low High Moderate

Consumption

Scalability High Very High Low High

Failure High Very High Low High

Tolerance

3.3 - Simulation Configuration

Here, we provide the simulation details for four UAV swarm communication approaches: hybrid
communication, decentralized communication, centralized communication, and distributed
communication with Kalman filtering. The simulations were performed using a custom Python
program to evaluate key performance metrics under controlled conditions. The results are
presented with an emphasis on experimental constants, variables, and modifications made to
the code for accurate simulations.

16
3.3.1 - Environment
- Dimensions: The 3D simulation space is defined by bounds [-100, 100] in the x, y, and z
directions.
- Number of UAVs: 4 UAVs were used in each simulation, each equipped with onboard
sensors for obstacle detection.
- Simulation Steps: 50 steps were executed for each approach.
- Communication Radius: 40 units, representing the range within which UAVs can directly
communicate.

3.3.2 - Experimental Constants

- Obstacle Detection Probability: 0.5 (50% chance for a UAV to detect an obstacle in a
step).
- Energy Consumption per Message: 0.1 units per data transmission.
- Relaying Efficiency: 100% (no packet loss during relaying).
- Periodic Synchronization Interval: 10 steps.

3.3.3 - Experimental Parameters (For Comparision Basis)

- Latency: Time taken for obstacle data to propagate through the swarm.
- Obstacle Sharing Efficiency: Proportion of obstacles shared across the swarm.
- Detection Accuracy: Ratio of correctly identified obstacles to total obstacles.
- Energy Consumption: Total energy spent on communication.
- Scalability Metric: Evaluates performance with increasing swarm size.

17
3.4 - Simulation Results
3.4.1 - Hybrid Communication (Proposed Solution)
Performance Metrics:
- Average Latency: 0.31seconds
- Obstacle Sharing Efficiency: 92.70%
- Detection Accuracy: 98.20%
- Energy Consumption: 20.10 units
- Scalability Metric: 645.20

3.4.2 - Decentralized Communication with Single Sensor

Performance Metrics:
- Average Latency: 0.12 seconds
- Obstacle Sharing Efficiency: N/A
- Detection Accuracy: 73.40%
- Energy Consumption: 7.50 units
- Scalability Metric: 820.00

3.4.3 - Centralized Communication

Performance Metrics:
- Average Latency: 0.45 seconds
- Obstacle Sharing Efficiency: 87.50%
- Detection Accuracy: 94.10%
- Energy Consumption: 25.00 units
- Scalability Metric: 450.00

3.4.4 - Distributed Communication with Kalman Filtering

Performance Metrics:
- Average Latency: 0.29 seconds
- Obstacle Sharing Efficiency: 89.30%
- Detection Accuracy: 96.80%
- Energy Consumption: 19.20 units
- Scalability Metric: 670.30

18
Communication Average Obstacle Detection Energy Scalability
Approach Latency (s) Sharing Accuracy Consumption Metric
Efficiency (%) (%) (units)

Hybrid 0.31 92.7 98.2 20.1 645.2

Communication

Decentralized 0.12 N/A 73.4 7.5 820

Communication
(Single Sensor)

Centralized 0.45 87.5 94.1 25 450

Communication

Distributed 0.29 89.3 96.8 19.2 670.3

Communication
(Kalman
Filtering)

3.4.5 - Environment Plots of Randomly Generated Objects during

Simulation Process

- Blue points: UAVs (random movement enabled)

- Yellow points: Partially detected obstacles (all UAVs not aware of presence of any
particular obstacle)
- Green points: Fully detected obstacles (all UAVs aware of presence of certain obstacle)
(The UAV behavior at every 5 steps with randomly generated obstacles)

19
20
4 - Conclusion
Hybrid Communication (Proposed Solution) proves to be the most well-rounded and effective
approach for UAV swarm navigation and obstacle communication. With the highest detection
accuracy (98.20%) and obstacle sharing efficiency (92.70%), this solution ensures superior
performance in obstacle detection and sharing, crucial for coordinated UAV operations in
complex environments. The energy consumption (20.10 units) is balanced, ensuring that the
system remains energy-efficient while delivering optimal performance. Furthermore, the
scalability metric of 645.20 indicates that this solution can effectively scale to larger UAV
swarms, maintaining robust communication and performance across multiple units.

In comparison to the other strategies, the hybrid communication approach strikes the best
balance between low latency, high accuracy, efficient obstacle sharing, and manageable energy
consumption. This makes it the ideal choice for real-world applications where both performance
and scalability are critical. Therefore, the hybrid communication solution stands out as the most
suitable and robust choice for advanced UAV swarm navigation systems.

21
5 - References
- Wang, J., Xu, W., & Yu, Z. (2019). "Decentralized Coordination of UAVs for Surveillance
Missions."
- Control-Aware trajectory predictions for Communication-Efficient Drone swarm
coordination in cluttered environments. (n.d.). Ar5iv.
https://ar5iv.labs.arxiv.org/html/2401.12852
- Yang, H., Zhang, Y., & Zhao, H. (2020). "Centralized Control of UAVs in Dynamic
Environments." IEEE Access.
- Control-Aware trajectory predictions for Communication-Efficient Drone swarm
coordination in cluttered environments. (n.d.). Ar5iv.
https://ar5iv.labs.arxiv.org/html/2401.12852
- Liu, X., Zhang, Y., & Wang, L. (2021). "Kalman Filtering for UAV Collision Avoidance in
Distributed Swarm Systems." Journal of Field Robotics.
- Control-Aware trajectory predictions for Communication-Efficient Drone swarm
coordination in cluttered environments. (n.d.-b). Ar5iv.
https://ar5iv.labs.arxiv.org/html/2401.12852

22
6 - Work Distribution

Group Member Assigned Work Presentation Section

ABHINAV REDDY KALLEM Research and Conclusion

Comparative Analysis

ABHIRAM JUTURU Performance Metrics / Midsem Summary

Report Documentation

ANMOL MURTI Grammar Check / Features and Constraints

Proofreading

CH V V SAI PRAJEET Performance Monitoring / Problem Statement

Data Analysis

ROHAN REDDY AMANAG… Algorithm Development / Proposed Solution

Performance Metrics

VENKATA TEJA PAPPU System Integration / Comparative Analysis

Validation

23