E-Motorbike Telemetry Dashboard System/h4>

- Developed an interactive dashboard using Python and Raspberry Pi 3 to monitor and analyze motor controller data via CAN (Controller Area Network) communication.

- Collected key motor parameters such as RPM and current supply from the motor controller over CAN.

- Implemented C code on ESP32 to act as a CAN receiver, filtering messages based on specific ID fields and forwarding relevant data via UART to the Raspberry Pi.

- Designed a graphical dashboard using Pygame on the Raspberry Pi, providing a real-time visual interface for motor performance monitoring.

- Integrated a data logging system and saved the motor rpm and current data in CSV format for analysis and diagnostics.

Smart Temperature Control for Hydroponic using DS18B20 temperature sensor and STM32

- Integrated an STM32 microcontroller, a DS18B20 temperature sensor via the 1-Wire protocol, and a relay-controlled fan to regulate temperature in a hydroponics system.

- Developed a PID-based control system to dynamically adjust the fan speed for maintaining optimal conditions.

- Implemented FreeRTOS with one task dedicated to reading temperature data and another for executing the PID control, ensuring real-time responsiveness and efficient temperature regulation.

IIOT WATCH THAT MEASURES PULSE

- Developed a smartwatch using ESP32 that integrates a SEN-1157 Pulse Sensor for real-time heart rate monitoring.

- Displayed heartbeat data on an SSD1306 OLED display using I2C communication.

- Implemented WiFi connectivity to fetch and display the current time and related information, ensuring seamless real-time updates.

PID-Based Flight Controller for Quadcopter

- Developed a flight controller for a quadcopter by integrating an MPU-6050 IMU sensor with an Arduino.

- Acquired gyroscope and accelerometer data from the MPU-6050 and processed it to compute roll, pitch, and yaw angles.

- Implemented a complementary filter to fuse accelerometer and gyroscope data, compensating for gyroscope drift and improving attitude estimation.

- Designed a PID-based flight controller to stabilize the quadcopter by adjusting motor speeds.

- Generated PWM signals to control ESCs, which regulated the BLDC motors for stable flight dynamics.

CUSTOM ROBOT NAVIGATION WITH NAV2

This project demonstrates custom robot navigation using ROS2 Humble and NAV2, incorporating SLAM for mapping and path planning. The robot utilizes Lidar sensor technology for accurate environmental perception.

TurtleBot3 Trajectory Recorder and Visualizer

This Project records the trajectory of a TurtleBot3 robot as it moves through its environment and visualize it in ROS2 Humble.

AUTONOMOUS DRONE BASED ON MOCAP

Designed and developed a drone flight controller using ESP32 integrated with Micro-ROS for real-time interfacing with a Motion Capture System, programmed in C and C++ using ESP-IDF.

- Implemented an RTOS-based scheduler to acquire Motion Capture System data and activate fail-safe mechanisms during communication loss.

- Integrated an MPU-6050 IMU (via I2C) and Lidar (via UART) to enable autonomous safe landing under fail-safe conditions.

PRESCRIBED PERFORMANCE CONTROL USING DC MOTOR

- Developed the embedded components for a Prescribed Performance Control (PPC) algorithm to regulate DC motor speed on a Teensy microcontroller, achieving 30% better stability compared to conventional PID control.

- Implemented the PPC formula in code and processed the control output into pulse signals to drive the DC motor’s ESC, ensuring precise speed regulation.

PAYLOAD SYSTEM FOR AUTONOMOUS DRONE

- Redesigned the payload system for an autonomous drone using CAD and 3D printing, integrating a Raspberry Pi, IMU, and battery for optimized weight distribution and stability.

- Fabricated the payload structure using carbon fiber for durability and lightweight performance.

- Designed and 3D-printed the casing for electronic components using SolidWorks and manufactured it with a Creality 3D printer.

- Assembled the electronic circuitry by soldering components onto a breadboard and incorporating switches to enable controlled power distribution and system isolation for safety and debugging.

Traffic Sign Classification Using Cross Stage Partials(CSP)

- Developed a traffic sign classification model utilizing CSPNet, which partitions the feature map to reduce computational load while maintaining efficiency. This approach improved gradient flow and computational efficiency by 20%.

- Implemented the Mish activation function, which enhances gradient propagation compared to ReLU, leading to better feature representation and smoother optimization.

- Applied data augmentation techniques to improve generalization and mitigate overfitting—common challenges in traffic sign classification.

- Achieved 87% accuracy on the German Traffic Sign Dataset.

A* Path Planning with PID Control on TurtleBot4

- Developed a path-planning system that loads and processes map data to construct graph structures for optimal path calculation using the A search algorithm*, a heuristic-driven approach that balances cost and efficiency.

- Implemented A search*, which combines Dijkstra’s algorithm (guaranteeing shortest paths) with a heuristic (Euclidean distance) to prioritize exploration, improving computational speed.

- Designed a PID controller to ensure accurate path following by dynamically adjusting the robot's velocity and orientation.

- Validated the approach on TurtleBot3 simulations and a physical TurtleBot4, demonstrating reliable real-world performance.

RRT* Path Planning and Following for Obstacle Avoidance on TurtleBot3 Simulation

- Implemented the RRT* (Rapidly-exploring Random Tree) algorithm for path planning, leveraging a KD-Tree for efficient nearest-neighbor searches and graph construction.

- RRT* works by randomly sampling points in the environment, extending a tree structure toward them, and iteratively refining paths by rewiring connections to minimize cost, ensuring asymptotic optimality.

- Used KD-Tree ( library implementation), a space-partitioning data structure, to efficiently store and retrieve nearest nodes, significantly improving the performance of path expansion.

- Designed a PID controller to track the generated path accurately by dynamically adjusting the robot's motion.

- Validated the approach through simulations in RViz and Gazebo, ensuring effective real-world applicability.

Ball-Following Turtlebot using Kalman Filter

- Developed a ball-following system for TurtleBot3 using a Kalman Filter to enhance tracking accuracy and predict future ball positions based on motion dynamics.

- Implemented ball detection using OpenCV, identifying a red ball and estimating its distance based on its perceived size—larger size indicating closer proximity.

- Utilized the Kalman Filter, which recursively estimates the ball’s position based on its current velocity and motion model and by combining noisy sensor data with these motion predictions it helps in improving tracking stability and thereby reducing errors.

Gesture Controlled Robotic Arm

This system enables real-time control of a robotic arm using hand gestures detected via MediaPipe. The gestures are processed in Python and translated into control commands, which are transmitted over UART to an STM32 microcontroller. The STM32 is programmed using CMSIS and HAL drivers, with hardware timers configured in PWM mode to control servo motors. Each timer channel generates pulse widths (typically 0.5–2.5 ms within a 20 ms period) corresponding to specific joint angles. The microcontroller continuously reads UART data, decodes the incoming gesture commands, and updates PWM outputs accordingly.

All these taks are controlled using FreeRTOS to ensure all the tasks are performed in a determinisitic manner.

This project demonstrates the integration of vision-based HMI with embedded motor control using timer-driven PWM and efficient serial communication.