Designing architectures at the edge of Fluid Mechanics and Artificial Intelligence.
"Small Scale Raspberry Pi Based Autonomous Car: A Practical Guide to Understand Complexity"
READ ABSTRACT
Proper Orthogonal Decomposition (POD) for turbulence prediction.
Leading Embedded Systems Architecture & Control Logic.
Standard web-based voting applications are vulnerable to Cross-Site Scripting (XSS) and database injection. The objective was to build a tamper-proof voting engine that operates in an isolated shell environment, eliminating surface area for web-based attacks.
THE ARCHITECTUREThe system is architected around a deterministic state machine written in pure C. It utilizes raw memory management (Structs) to handle voter sessions and candidate registries. To ensure auditability, the system implements a custom File I/O persistence layer. Every vote cast is serialized and locked into a binary file, ensuring that data survives power cycles and program termination without relying on heavy external SQL databases.
ACCESS REPOSITORYIndustrial machinery often operates in remote agricultural fields where Wi-Fi and Bluetooth connectivity are non-existent. Traditional automation solutions fail in these off-grid environments.
THE SOLUTIONThis architecture utilizes the A7670C GSM module to establish a telemetry link over the cellular network. The system parses incoming SMS text strings using a custom command interpreter on the microcontroller. It validates the command syntax and triggers high-voltage relays to actuate pumps or motors. This approach guarantees control capability over infinite distances, provided there is basic cellular signal.
ACCESS REPOSITORYThe system interfaces with a MAX30102 oximetry sensor to capture raw photoplethysmogram (PPG) data. A microcontroller processes these analog signals to filter out noise and compute Heart Rate and SpO2 levels in real-time.
CLOUD TELEMETRYInstead of local storage, the processed vitals are serialized into lightweight packets and transmitted via the ESP8266 Wi-Fi module to the Blynk IoT Cloud. This architecture allows medical professionals to visualize patient status on a remote dashboard with sub-second latency, facilitating immediate response to critical events.
ACCESS REPOSITORYStandard hobbyist robots use "blocking" delays, causing the processor to freeze during movement, leading to jerky controls. I implemented a non-blocking asynchronous control loop using `millis()`. This allows the processor to handle motor PWM generation and Bluetooth UART communication simultaneously.
VELOCITY PROFILINGTo prevent mechanical wear and ensure smooth motion, I developed a "Soft-Start" algorithm. Instead of jumping from 0 to 100% power instantly, the firmware ramps up the Pulse Width Modulation (PWM) duty cycle linearly over time intervals. This creates a fluid velocity profile similar to industrial AGVs and enhances stability.
ACCESS REPOSITORYThis work explores the architectural complexities of building small-scale autonomous vehicles using Raspberry Pi. It provides a practical guide for engineers to understand the interplay between hardware constraints and software efficiency, specifically focusing on TCP/IP protocols and real-time communication libraries.
AUTHORSNallabati Venkata Durga Sai Sarat Chandra, Nunna Mokshith, Asha C A, Sreekumar P. C
TECHNICAL SCOPEThe paper reviews Python TCP/IP libraries (socket, asyncio, Twisted, Scapy) and their implementation in differential-drive mobile robots. It discusses ethology-inspired action selection mechanisms and the integration of ultrasonic sensor arrays for precision distance measurement in automotive prototypes.
VIEW ON RESEARCHGATESolving the non-linear Navier-Stokes equations for turbulent flows via Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES) is computationally prohibitive. High-fidelity simulations can take hours or days to converge, creating a major bottleneck for design optimization cycles and real-time control applications.
METHODOLOGY: PROPER ORTHOGONAL DECOMPOSITION (POD)This research leverages Proper Orthogonal Decomposition (POD) to extract the dominant coherent structures (energy-containing modes) from flow field data. By performing Singular Value Decomposition (SVD) on the snapshot matrix, we identify a low-dimensional subspace that captures the majority of the flow's kinetic energy.
REDUCED ORDER MODELING (ROM)We construct Reduced Order Models (ROMs) by projecting the governing equations onto these POD modes. The result is a surrogate model that retains the essential physics while reducing the degrees of freedom from millions to just a few dozen. This achieves computational speedups of O(10^3), enabling near-instantaneous flow prediction suitable for Digital Twins and active flow control.
ACCESS REPOSITORYAs the Head Engineer for the 'Pasupatha' department within Club Mirai, I lead the design and implementation of the entire embedded systems architecture for our companion robots. This involves overseeing the integration of sensors (LiDAR, IMU, Ultrasonic), power distribution units, and microcontrollers.
TECHNICAL SCOPEMy responsibility extends beyond simple kinematics. I architect the communication bus (I2C/UART/CAN) that links the high-level ROS navigation stack (running on SBCs like Raspberry Pi/Jetson) with the low-level motor controllers (Arduino/STM32). I ensure synchronization between the navigation goals and the physical actuation of the robot.
ACCESS REPOSITORYPresented the research paper titled "Complexity in Autonomy" at the International Conference on Robotics and Mechatronics (ICRM) 2025. The presentation focused on the architectural trade-offs in edge computing for small-scale autonomous agents.
IMPACTThe work highlighted the critical "blind time" in single-threaded robotic vision systems and proposed a novel multi-threaded solution. Engaged with industry experts on the future of real-time operating systems (RTOS) in consumer robotics.
VIEW PUBLICATIONThe hackathon presented complex problems regarding quantum circuit optimization and state preparation using IBM's Qiskit framework.
THE SOLUTIONI successfully navigated the challenge by integrating Generative AI (LLMs) into the workflow. By leveraging AI to generate boilerplate Qiskit code and optimize gate sequences, I was able to focus on the high-level quantum logic, solving the hackathon problem statement efficiently.
Participated in the ASME Engineering Festival (EFx) held at CBIT Hyderabad. Engaged in mechanical design challenges and rapid prototyping workshops.
ACCESS REPOSITORYOrchestrated the creation of digital assets for non-profit campaigns, focusing on fundraising and public outreach. Designed thematic posters and dynamic video reels that significantly boosted user engagement on social media channels.
DESIGN WORKFLOWEmployed a professional toolchain centering on Adobe Photoshop for complex image manipulation. Integrated AI-driven design tools to optimize workflow efficiency and create compelling visual narratives that resonate with donor audiences.
ACCESS REPOSITORY
MAJOR SYSTEMS UNDER CONSTRUCTION.
REASON: Academic Clearance Level 4/8.
Deep-system integration requires advanced semester completion. Current focus remains on foundational mastery and minor system modules.