Projects in SidlaghattaProjects in Sidlaghatta
Cover image for Title: FlowSense Pro  -
Title: FlowSense Pro - Industrial Flow & Chemical Dosing SCADA Gateway with OPC-UA & BLE Problem: Municipal wastewater treatment plants, breweries, and chemical processing facilities rely on analog 4–20 mA pressure/flow transmitters and mechanical pulse flow meters that lack native network connectivity. Modernizing these field instruments into SCADA platforms historically required dedicated automation racks and programmable logic controllers, inflating per-station deployment costs. Additionally, field technicians in wet, hazardous environments struggle with device commissioning and routine calibration because older systems require physical laptops, terminal blocks, and USB-to-serial cables. Traditional field gateways also lack secure, fail-safe remote firmware updates (OTA), risking operational downtime during software maintenance. Solution: Built a ruggedized, DIN-rail mounted smart gateway around the ESP32-S3 (Xtensa dual-core 240MHz) utilizing ESP-IDF v5.x: High-Precision Dual Measurement: Combines a 16-bit ADS1115 ADC and a 0.1% precision 100 Ω shunt to read 4–20 mA current loops with ±0.5% calibrated accuracy, alongside a hardware pulse counter (PCNT) interrupt engine reading turbine flow meters up to 500 L/min. Direct SCADA & OPC-UA Integration: Compiles the open-source open62541 OPC-UA stack directly into ESP-IDF, enabling the gateway to communicate natively with SCADA servers and industrial HMIs over TCP/IP without intermediate PLC hardware. Cable-Free BLE Commissioning: Implements a custom BLE 5.0 GATT profile that allows field operators to calibrate sensor K-factors, set alarm thresholds, and configure local Wi-Fi credentials via an iOS/Android mobile companion app in seconds. Zero-Power Display & Fail-Safe Actuation: Integrates an ultra-low-power 2.9" SPI E-Paper display that maintains critical process readouts even during brownouts, alongside four opto-isolated 24V/3.5A power MOSFET outputs driving dosing pumps and emergency shutoff valves in <10 ms. Enterprise Security & Robust OTA: Employs mutual-TLS (mTLS) certificate-based authentication with AWS IoT Core and dual-partition HTTPS OTA with automated bootloader rollback upon boot failure.
0
16
Cover image for Title: SmartMill Guardian - CNC
Title: SmartMill Guardian - CNC Machine Predictive Health & Vibration Diagnostics Edge System Problem statement: High-precision CNC milling centers and industrial machining tools frequently suffer from sudden spindle bearing degradation and thermal runaway. A single unplanned spindle seizure costs manufacturing facilities upwards of $40,000 per incident in damaged workpieces, emergency tooling, and shop-floor downtime. Traditional plant maintenance relies on manual spot checks every 4 to 8 hours or cloud-only IoT gateways that suffer from 2–5 second network latencies—far too slow to trigger an emergency feed-hold before catastrophic mechanical failure. Furthermore, retrofitting older factory machinery with proprietary PLC hardware (e.g., Siemens S7 racks) is prohibitively expensive and requires disruptive rewiring. Solution: Designed and deployed a deterministic edge monitoring unit powered by a Raspberry Pi 4B running bare-metal FreeRTOS and Embedded C (C99): High-Rate Vibration FFT: Interfaces with an ADXL345 3-axis MEMS accelerometer via hardware SPI at 1 kHz ODR. An on-chip CMSIS-DSP Fast Fourier Transform (FFT) paired with a lightweight µTensor INT8 model identifies bearing defect frequencies (BPFO/BPFI harmonics) with 94% accuracy. Thermal Profiling & PLC Bridging: Continuously monitors bearing temperatures using an MLX90614 infrared sensor over I²C while polling spindle speed, drive current, and status words from existing Siemens S7-1200 PLCs over RS-485 Modbus RTU. Deterministic Emergency Interlock: Features a 4-channel opto-isolated relay interface capable of triggering a CNC machine feed-hold or spindle stop in <5 ms upon anomaly detection—eliminating cloud latency. Cloud & SCADA Telemetry: Securely streams time-series vibration metrics and alerts via TLS 1.3 MQTT to AWS IoT Core, populating live shop-floor InfluxDB and Grafana dashboards for condition-based maintenance scheduling.
0
24
Cover image for Title: Smart Assistive Gesture Glove
Title: Smart Assistive Gesture Glove and Caretaker Alert System 1. Problem statement: Speech-impaired, partially paralyzed, or bedridden patients often struggle to communicate basic necessities (food, water, washroom, physical discomfort) or call for immediate emergency help without a caregiver continuously in the room. Standard push-button call systems lack expressive nuance, while cloud-reliant smart devices fail completely whenever local Wi-Fi or internet connectivity drops, posing a critical safety hazard during medical emergencies. 2. Solution and Technical Implementation: ​Gesture-to-Intent Recognition: The patient wears a wearable glove equipped with four flex sensors. Using trimmed-mean signal sampling and Exponential Moving Average (EMA) filtering, the system filters out muscle tremors and maps individual finger flexes to primary needs (Food, Washroom, Water, Discomfort), with a chorded fist gesture reserved for an instant Emergency SOS. ​Zero-Downtime Local Transmission: Communication between the glove and the caregiver base station uses peer-to-peer ESP-NOW radio protocol. This operates completely independent of routers or internet access, featuring an auto-channel hopping algorithm on the glove to maintain the radio link even if radio channels shift. ​Multi-Modal Feedback & Cloud Alerts: The receiver base station converts incoming packets into immediate spoken voice alerts via a DFPlayer Mini MP3 module, visually displays the request on an SSD1306 OLED, and sounds an active buzzer during SOS events. ​Non-Blocking Remote Gateway: An asynchronous FreeRTOS background task pinned to Core 0 relays every command and glove-disconnection warning directly to the caregiver’s smartphone via a Telegram Bot, ensuring family members stay informed even when away from the base station without delaying real-time local alarms.
0
32