1.       Introduction

Autonomous indoor navigation represents a fundamental challenge in the field of mobile robotics, with applications extending across domains such as warehouse automation, healthcare, and domestic assistance. These systems aim to minimize human involvement and enhance operational efficiency by autonomously performing tasks like item delivery, cleaning, and surveillance [1]. A crucial component enabling such autonomy is Simultaneous Localization and Mapping (SLAM), which allows a robot to construct a map of an unknown environment while continuously determining its own position within that map [2]. Recent advancements have led to the widespread adoption of autonomous indoor robots across industries including logistics, healthcare, and personal service sectors. Earlier localization systems typically depended on fixed infrastructure such as RFID tags, ultrasonic beacons, or visual markers but contemporary approaches increasingly rely on map-based localization techniques

Fig. 1. Obstacle Detection Concept Overview

For example, 2D LiDAR - based mapping can be used to align sensor scans with occupancy grid maps to estimate the robot’s pose. Likewise, vision based methods have gained traction in recent years. Bajpai and Amir-Mohammadian (2021), for example, demonstrated real-time 3D indoor map- ping and navigation using the iPhone’s ARK it framework [4]. In this project, we present the design and implementation of a cost-effective autonomous indoor mobile robot developed on a Raspberry Pi 4 platform running ROS 2. The robot is equipped with a TF Mini LiDAR and ultrasonic sensors for depth perception and obstacle detection Fig 1,[5], [6]. It follows a conventional SLAM - driven navigation architecture (sense→ localize/map → plan → act) and leverages open - source ROS 2 modules such as Grid-based Mapping (gmapping) and Navigation. Visualization and monitoring are performed through RViz. The system is tested in a controlled laboratory environment, where the robot autonomously explores, maps its surroundings, and navigates between predefined goal points using only onboard computation and sensors, without reliance on GPS or external tracking systems [7]

Related Work

Autonomous indoor navigation has been the focus of extensive research within the robotics community. Numerous studies have examined mapping, localization, and motion- planning strategies for robots operating in structured indoor environments. Among these approaches, Simultaneous Localization and Mapping (SLAM) remains a fundamental technique that enables a robot to construct and update a map of its surroundings while continuously estimating its pose [8]. Most modern SLAM frameworks employ LiDAR or camera- based sensors due to their accuracy, reliability, and suitability for real-time applications. Early localization systems primarily relied on landmark- or beacon-based approaches, including RFID transmitters, ultrasonic beacons, and fiducial markers [9]. Although these systems provided stable position estimates, they required pre-installed environmental infrastructure, which limited flexibility.

Fig. 2. Path Planning Concept

In contrast, map-based localization techniques use onboard sensing and computational resources. For instance, 2D Li- DAR scan matching allows the robot to compare current observations with stored occupancy grids to estimate position. In practical implementations, we find that combining wheel odometry with LiDAR or visual data through probabilistic filtering-such as the particle filter or graph-based SLAM- improves robustness and accuracy. Recent advancements have also expanded the use of vision-based SLAM [9]. Systems based on monocular or RGB-D cameras can generate dense environmental models for navigation and augmented reality. Bajpai and Amir-Mohammadian (2021), for example, presented a markerless indoor mapping approach using Apple’s ARKit, where smartphones captured 3D spatial maps for shared localization among multiple devices. While promising, such solutions depend on high-performance mobile hardware and are not directly optimized for embedded robotic platforms [9], [10].

In our work, we emphasize LiDAR-based SLAM for its robustness and low computational demand. We use a planar TF Mini LiDAR to perform two-dimensional mapping in a ROS 2 environment [10]. This design choice simplifies real-time processing and ensures consistent performance under varying lighting conditions. For path planning, we examined both classical and sampling-based algorithms. Algorithms such as Dijkstra’s and A* provide deterministic, optimal paths in grid- based maps, while Rapidly-Exploring Random Trees (RRT) offer faster exploration in larger or higher-dimensional spaces. As highlighted by Jayaparvathy et al. RRT can outperform traditional methods in open environments due to its ability to rapidly sample feasible trajectories. In our system, we utilize the Navigation2 (Nav2) framework of ROS 2, which supports both global path planners (A* and RRT) and local controllers such as the Dynamic Window Approach (DWA) [11].

Obstacle Avoidance: Obstacle avoidance: It is another essential component of our navigation pipeline. As supported by previous studies, range sensors-particularly LiDAR and ultrasonic modules-are effective for detecting nearby obstacles and ensuring collision-free motion. We adopt a sensor-fusion approach, where LiDAR provides precise distance estimation and ultrasonic sensors supplement perception in cases where transparent or reflective surfaces might degrade LiDAR performance [11]. Cameras may also be used for detecting obstacles or visual landmarks, but we find they often demand greater computational resources and are more sensitive to environmental lighting Fig 3.

Fig. 3. Obstacle Avoidance

Through this hybrid sensing strategy, we ensure safe and reliable motion even in cluttered or dynamic indoor environments. Our review of existing literature indicates that contemporary indoor robots primarily rely on LiDAR-based SLAM, sensor fusion, and path planners such as A* and RRT for efficient navigation. Building upon these established frameworks, we implement and validate a ROS 2–based navigation system that integrates SLAM Toolbox and Nav2 modules on a TurtleBot4 platform, demonstrating effective mapping, localization, and path execution in real-world conditions [12].

2.       Components And Architecture

This section describes the hardware components we used for building our Smart Autonomous Indoor Navigation Robot. Each component has its contribution to the sensing, control, power management, and motion of the robot. The system architecture integrates all these modules within a single frame- work that guarantees reliable and smart indoor navigation.

Fig. 4. TF-Mini-S Micro LiDAR Module

A.         TF-Mini-S Micro LiDAR Module

The TF-Mini-S LiDAR sensor is a compact, low-cost, and high-performance sensor for precise distance measurement Fig 4. It works on the principle of Time-of-Flight (ToF), which calculates the distance based on the time a pulse takes to travel to an object and back. The module offers stable and accurate measurement and communicates via UART or I²C interface, making it suitable for embedded systems and robotics. In this project the TF-Mini-S LiDAR module is used as the main obstacle detection sensor, which enables accurate mapping of the surrounding environment for intelligent navigation decisions to avoid collisions during autonomous operation Table 1.

Table 1: Specifications of different Lidar modules

B.        

Fig. 5. Raspberry Pi 4 (Model B)

The Raspberry Pi 4 Model B is a powerful microcomputer with a 1.5 GHz quad-core ARM Cortex-A72 processor and has memory ranging from 2 GB to 8 GB LPDDR4 RAM. It supports full operating systems such as Raspberry Pi OS or Ubuntu and includes versatile connectivity options like USB 3.0, HDMI, and Ethernet Fig 5. It make Rpi more suitable for computational resources such as image processing and real- time navigation algorithms. In our project, the Raspberry Pi 4 serves as the CPU as it manages high-level decision-making, path planning, and sensor data fusion from LiDAR and IMU modules for intelligent autonomous movements.

C.       Arduino UNO R3

Fig. 6. Arduino UNO R3)

D.      

Fig. 7. L298N Motor Driver

The L298N is a dual H-bridge driver having the capability of driving two DC motors, up to 2 A per channel, and handling voltages from 5 to 35 V Fig 7. It allows bidirectional control of motor rotation, and it supports PWM-based speed control. In our autonomous robot, the L298N module interfaces with Arduino UNO for regulation of speed and direction of the motors, allowing precise movements in all directions and also turning maneuvers required for smooth indoor navigation.

E.      

Fig. 8. LM2596 DC-DC Buck Converter

The LM2596 is a high-efficiency step-down voltage regulator that converts higher DC voltages in the neighborhood of 40 V down to much lower, producing stable outputs, such as 5 V or 3.3 V Fig 8. As it has high conversion efficiency and low thermal output, it is well suited for multi-voltage components, which prevents voltage fluctuations and maintains consistent system performance Fig 9.

Fig. 9. Component Architecture.

3.       Methodology

The robot navigation pipeline operates in real time in a loop of sensing, localization/mapping, planning, and control:

1.     Sensor Data Collection: Data from sensors is continuously received by the robot. Indeed, TF Mini Li- DAR produces 2D range scans, that is to say, arrays of distance measurements. Ultrasonic sensors report close-range distances. Wheel encoders and IMU provide odometry, thus, incremental motion estimates. All raw data is published on ROS2 topics. For instance, the Li- DAR node publishes a LaserScan message with obstacle distances around the robot.

2.     Localization and Mapping (SLAM): SLAM solves both localization and mapping problems in a coupled manner. In the context of this work, we consider a filter-based SLAM approach, namely Rao Blackwellized Particle Filter or Extended Kalman Filter. Each incoming LiDAR scan is matched to the current map in order to estimate the robot’s pose. In practice, we rely on the SLAM Toolbox package, which internally uses scan matching and a particle filter in order to update both the robot’s pose and the map. Similarly to standard SLAM, the localization step compares the new scan against the map, e.g., an occupancy grid, in order to refine the pose. On the other hand, the mapping step updates the occupancy grid given newly observed free or occupied cells. One result is a progressively built map in 2D about the environment. Once the map is built-after some exploration-we can switch, if needed, to a localization- only mode, for example, AMCL, using the fixed map.

3.     Path Planning: The planner computes a collision-free path given a start, i.e., the current pose, and a goal coordinate.

4.     Motion Control and Execution: The planned path is converted into velocity commands. A simple feedback controller, say PID on heading and distance, converts these waypoints to linear and angular velocities, published to the level topic. The robot executes the first segment of the path for a short duration, then the loop repeats: it senses again, localizes and replans as necessary. It allows dynamic re-planning in case unexpected obstacles appear. If there is an obstacle detected on the immediate path, by either LiDAR or ultrasound, the robot may stop and replan a new path. Alternatively, it could use a reactive avoidance rule, such as sidestepping. Throughout, rviz2 displays the evolving map, the robot’s pose, and the planned trajectory to monitor.

Fig. 10. Flow Chart

4.       Result And Discussion

The Smart Autonomous Indoor Navigation Robot was successfully designed, implemented and tested in controlled indoor environments to evaluate its performance in obstacle detection, path navigation and autonomous decision-making in different conditions. Our proposed Robot display how well we have integrated hardware into the system, with Raspberry Pi 4 communicating with Arduino UNO, and associated sensor and actuators [13]. The TF-Mini-S LiDAR module presented accurate distance measurements within a range of 0.1-12 m with error below ±4 cm. The high accuracy allowed the robot to detect obstacles and change its route in real time.

During navigation testing, we placed the robot in difficult indoor layouts containing both static and dynamic obstacles. The pair of LiDAR and ultrasonic sensor made sure that effectiveness of multi-sensor fusion was not lost by enabling the robot to perceive both far- and near range objects accurately. The MPU-6050 IMU enhances stability and orientation control, by maintaining its balance, which allows smoother turns and reduces drift.

Fig. 11. SLAM-based mapping

Fig. 12. Prototype Image I

The RS-775 DC motors, controlled by the L298N motor driver, ensured reliable motion with adequate torque on indoor surfaces. The LM2596 DC-DC buck converter maintained stable voltage level for different integrated modules, which helps in consistent performance throughout extended operation periods. The active buzzer helped in providing audible feedback related to operational states like power-up, proximity to obstacles, and task completion [14]. We found our system to be cost-effective and efficient, with integrated hardware and flexible software architecture for further extensions. The result justify the feasibility of an affordable autonomous navigation platform capable of carrying out indoor mapping, collision avoidance, and path optimization with less human supervision. The system’s successful operation and performances under real-time condition indicates its applicability in various application from home automation, warehouse logistics and various indoor service robotics.

5.       Conclusions:

This research work focused on the design and implementation of a smart autonomous indoor navigation robot that could perform effective navigation and obstacle avoidance within confined indoor environments.

Fig. 13. Prototype Image II

The proposed project integrates various sensors and control modules together for autonomous operation, such as high-precision distance mapping using TF- Mini-S LiDAR along with proximity sensing through HC- SR04 ultrasonic sensors and motion detection through an MPU-6050 [15]. Data fusion and decision processing are handled by Raspberry Pi 4, while Arduino UNO R3 does low- level motor control through the L298N motor driver and RS- 775 motors to achieve stable and synchronized movement. The experimental results prove that the robot can detect obstacles effectively, maintain balance, and move smoothly in indoor spaces. The integrated LM2596 buck converter provides stable power to all components without any fluctuations, and the active buzzer helps in system feedback for better safety and alerting [16], [17].

Future improvements will be focused on improving the autonomy and perception of the robot using SLAM techniques, object detection based on computer vision, and decision systems based on machine learning. These enhancements could further increase the efficiency, adaptability, and intelligence of our proposed indoor navigation robot by making it suitable for a wide range of domestic, industrial, and research applications [18].

6.       Future Scope

The development of the Smart Autonomous Indoor Navigation Robot lays a strong foundation for further research and enhancement in autonomous systems. Although the robot prototype we presented is capable of successful and reliable indoor navigation with obstacle avoidance, in the future, further enhancements can be made to enhance its capabilities. The SLAM technology allows the robot to build and develop dynamic maps of its environments, therefore increasing its spatial awareness and its localization accuracy. With the use of cameras and advanced image processing techniques, computer vision systems can allow the robot to recognize objects, signs, and dynamic obstacles, improving decision-making in the case of complicated indoor spaces. In addition, we can infuse machine learning algorithms for optimized path planning as the robot learns from previous navigation experiences to adapt to changes in environmental conditions. Future development of the system might include cloud connectivity and IoT integration, such as remote monitoring, real-time data analysis, and multi-robot coordination. Improving power efficiency and precision of motor control for enhanced operational stability and reliability. All these improvements will allow the robot to perform tasks in a wider range of applications, including ware- house automation, indoor delivery, surveillance, and assistive robotics in calculative environments.