This book presents analysis of various intelligent approaches, including applicable heuristic graph theoretic and bio inspired techniques for achieving optimal point-to-point navigation. Usage of LiDAR and RGB-D sensor data as source input has been preferred, ensuring total workspace coverage and minimization of action performed by the robot in carrying out realistic applications in certain congested environment. Sampling-based approaches which use arbitrary information gain formulation and Learning-based techniques, both are studied extensively. Uniform Sampling-based techniques are found feasible in exploring the state space without any complexity in geometrically modeling the configuration area ensuring embedded intelligence into the mobile robots in finding optimal execution. The navigation over both static as well as dynamic obstacles are analysed and the observations are presented in a comparative manner.
For dynamic environment, it is somewhat comparatively difficult for achieving proper path navigation. VSLAM (Visual Simultaneous Localization And Mapping) uses the data captured by externally perceived sensors for the purpose of self-locating and simultaneous map-building leading to understanding the unknown environment. This thesis also proposes a keen way to detect onroute obstacles using training of model through adversarial neural network along with 3D reconstruction of a concerned surrounding followed by memory tracing of already explored path by the mobile agent for ease in achievement of optimized path from start to desired goal position. In case of GPS-denied indoor environment primarily the robot works based on its first hand sensor data, for example, proximity analysis, distance measure etc. In various scientific works it is observed that indoor robots face not only constraint space challenge but also systematic maneuver, path planning and path finding in case of cluttered environment.
Primary contributions of the work include LiDAR data inference by 2D Hect SLAM, Construction of Fusion SLAM accumulating 2D and 3D depth features and Geometric Optimization of navigation planning algorithms. The thesis concludes with the graphical and numerical analysis of the accuracy achieved using mentioned algorithms and specific benchmarking of the performance of used techniques.
Contents
1 Significance of Reactive navigation with motion plan for Point-to-Point mobile robots
1.1 Introduction
1.2 Background of the work
1.2.1 Perspective Localization of objects
1.2.2 Understanding Point-to-point navigation
1.2.3 Basic principle of Reactive Navigation
1.2.4 Map based motion plan
1.2.5 Need of research in Path planning strategies for mobile robots
1.3 Research questions, objectives and identified research gap
1.4 Research Framework and Hypothesis
1.4.1 Experimental Platform Design
1.4.2 Multi Sensor Fusion
1.4.3 Visualizer Option
1.4.4 Real-time map based planning
1.4.5 Perception of indoor environment accompanied with obstacle localization
1.4.6 Use of path planning approaches
1.5 Thesis Contribution
1.5.1 LiDAR data inference by 2D Hect SLAM
1.5.2 Construction of Fusion SLAM accumulating 2D and 3D depth features
1.5.3 Geometric Optimization of navigation planning algorithms
1.6 Organization of the Thesis
2 Literature Review
2.1 Introduction
2.2 Review Method
2.3 Localization and mapping followed by path planning accompanied with Visualization
2.4 SLAM mapping approaches followed by static and dynamic obstacle detection for proper path navigation
2.5 Application of Obstacle Detection Algorithm for executing smooth navigation
2.6 Introduction and application of path planning strategies for robot navigation
2.7 Introduction of Reactive navigation and application
2.8 Optimization in path finding
3 Visual Representation of mobile robot trajectory and environment recreation
3.1 Functional analysis of SLAM
3.1.1 Features of 2D Hect SLAM
3.1.2 Features of 3D depth SLAM
3.2 Depth Estimation Approach
3.3 Obstacle Estimation and Mapping Approach
3.3.1 Pose Estimation
3.3.2 2D Hect Mapping
3.3.3 Point cloud mapping
3.4 3D reconstruction of indoor environment with SLAM Data
3.5 Analysis of mapping algorithms
4 On path Object Detection and Recognition for collision free navigation
4.1 General ideas of object Detection and Recognition
4.2 Derivation of obstacles from objects
4.2.1 Double stage Object Detection (DSOD)
4.2.2 Single-stage Object Detection (SSOD)
4.2.3 YOLO v4 in on route obstacle detection
4.2.4 YOLO v5 in Microbial Object Detection (MOD)
5 Fundamentals of Mobile robot Navigation
5.1 Introduction
5.2 Map based path plan
5.3 Analysis of path plan Algorithms
6 Experimental Methodology for executing path planning strategy
6.1 Experimental setup
6.1.1 Workstation setup
6.1.2 Camera calibration
6.2 Stepwise Methodology of experiment set up
6.2.1 Integration of RPLidar with ROS
6.2.2 Integration of Realsense with ROS
6.2.3 Integration of Arduino with mobile robot for navigation
7 Data Analysis and validate experiment
7.1 SLAM results
7.1.1 2D Hect results
7.1.2 3D depth
7.2 Results obtained from sensor fusion accompanied with SSOD
7.3 Reactive navigation results
7.3.1 Graph Theoretic approach results
7.3.2 Bio inspired approach results
8 Discussions and Conclusion
8.1 Discussion
8.2 Summary of Contribution
8.3 Limitations of the Research and Scope of Future Work
8.3.1 Suggestions for the Next Researcher
Research Objectives and Topics
This thesis investigates intelligent navigation strategies for mobile robots operating in GPS-denied indoor environments, focusing on achieving collision-free point-to-point movement. The primary research objective is to develop a robust methodology that integrates sensor fusion (LiDAR and RGB-D data), advanced mapping (SLAM), and object detection via machine learning to facilitate optimized path planning in cluttered areas.
- Development of F-SLAM (Fusion SLAM) by combining 2D and 3D depth features.
- Implementation of object detection algorithms (YOLO v4/v5) for real-time obstacle recognition.
- Analysis of graph-theoretic and bio-inspired search strategies for optimal trajectory determination.
- Development of a Geometric Optimization approach for refining navigation paths.
- Experimental validation on a customized, two-wheeled differential drive mobile robot platform.
Excerpt from the Book
1.2.2 Understanding Point-to-point navigation
Point-to-point robot is mainly applicable for accomplishing different works in indoor environment. This work uses a customized mobile robot equipped with depth sensors which works on the principle of differential drive. Kinematic equations of custom robot model (Cubot) are included in equation7.5,7.6,1.9,1.6,1.7,1.8. x˙ r(t) and y˙ r(t) (derivative of xr(t) and yr(t) respectively) represent the position of center of the mobile robot at an instant t. ψ˙ r(t) is the angle between robot head and the x-axis of global coordinate frame, v(t) and ω(t) are the forward linear velocity and angular velocity respectively.
x˙ r(t)=(v(t) + δv(t))cos(ψr(t)) (1.3)
y˙ r(t)=(v(t) + δv(t))sin(ψr(t)) (1.4)
ψ˙ r(t) = ω(t) + δω(t) (1.5)
After conversion of continuous time to discrete time, taking sampling time ΔT the equations of Euler method is represented as:
xr k+1 = xr k + (vk + δvk)ΔT cos(ψr k) (1.6)
yr k+1 = yr k + (vk + δvk)ΔT sin(ψr k) (1.7)
ψr k+1 = ψr k + (ωk + δωk)Δ (1.8)
(xr k,yr k,ψr k) denotes the robot position at time step k. vk and ωk are the linear and angular velocity respectively. δvk and δωk are the discrete time velocity noises and angular velocity noises respectively.
Summary of Chapters
1 Significance of Reactive navigation with motion plan for Point-to-Point mobile robots: Defines the research objectives, background, and framework for reactive navigation in indoor mobile robotics.
2 Literature Review: Provides a comprehensive survey of existing studies on SLAM, path planning, and object detection techniques.
3 Visual Representation of mobile robot trajectory and environment recreation: Details the SLAM modules used for indoor environment mapping and trajectory visualization.
4 On path Object Detection and Recognition for collision free navigation: Evaluates machine learning models, specifically YOLO variants, to identify and recognize static and dynamic obstacles.
5 Fundamentals of Mobile robot Navigation: Explores informed and uninformed graph-theoretic search algorithms for path finding.
6 Experimental Methodology for executing path planning strategy: Describes the hardware configuration and sensor calibration processes used in the experiment.
7 Data Analysis and validate experiment: Presents and analyzes experimental results, validating the efficacy of the proposed sensor fusion and navigation algorithms.
8 Discussions and Conclusion: Summarizes research findings and outlines the limitations and future research potential.
Keywords
Point-to-point navigation, Adversarial neural network, Heuristic algorithms, GPS-denied environment, Path planning, VSLAM, Sampling-based technique, Learning-based technique, Geometrical Optimization, SLAM, LiDAR, Sensor fusion, ROS, Mobile robots
Frequently Asked Questions
What is the core focus of this research?
The research focuses on enabling autonomous mobile robots to perform collision-free point-to-point navigation in unknown, GPS-denied indoor environments.
What are the primary thematic areas?
The work integrates Simultaneous Localization and Mapping (SLAM), sensor fusion using LiDAR and RGB-D cameras, machine learning-based object detection, and geometric optimization for path planning.
What is the primary objective of this thesis?
The primary goal is to establish an intelligent, optimized navigation framework that dynamically prioritizes obstacles and calculates safe trajectories in cluttered environments.
Which scientific methodology is predominantly applied?
The research employs a hybrid approach, combining sampling-based mapping techniques, deep learning for object recognition (YOLO), and heuristic search algorithms (A*, RRT, etc.) for path planning.
How is the main body of the work structured?
The work is organized into modules covering visualization/mapping, machine learning-based obstacle detection, and the evaluation of path planning algorithms, finalized by comprehensive data analysis.
Which keywords categorize this work?
The work is characterized by terms such as Reactive Navigation, VSLAM, LiDAR, Sensor Fusion, YOLO, and Geometric Optimization.
How does the proposed F-SLAM improve navigation?
F-SLAM improves performance by combining the specific environmental perception strengths of both 2D Hector SLAM and 3D RGB-D SLAM, resulting in more accurate collision avoidance data.
What is the significance of the "Zone Segmentation" approach?
Zone segmentation allows the robot to dynamically prioritize obstacles based on their distance and movement patterns, enabling more intelligent and efficient navigation decisions.
- Citar trabajo
- Rapti Chaudhuri (Autor), 2022, Reactive Navigation with Sensor Fusion and Geometric Optimization for Map-Based Path Planning. A Systematic Point-to-Point Navigation for AGV, Múnich, GRIN Verlag, https://www.grin.com/document/1247164
