App Development
BLE Sense (Android App)
BLE Sense is a comprehensive application that leverages Bluetooth Low Energy (BLE) technology to enable low-power, efficient communication between Android devices and BLE peripherals. Designed to meet the evolving needs of modern IoT ecosystems, BLE Sense excels at real-time data exchange with minimal latency.
Year :
2024-25
Industry :
Sensors
Client :
Annam.ai, IIT Ropar
Project Duration :
15 weeks
Problem :
Interfacing with multiple distinct IoT sensors typically requires separate proprietary applications and constant manual device pairing, which is slow and consumes significant battery. Furthermore, checking soil nutrient levels (NPK and Organic Carbon) traditionally relies on expensive laboratory testing or cloud-dependent analysis, making on-the-spot agricultural diagnostics impossible. Additionally, DIY robot controllers often lack real-time telemetry display, and there is a general lack of interactive, gamified ways to teach students how Bluetooth signal propagation and proximity sensing work in the real world.
Solution :
We developed BLE Sense, a unified, native Android application built with Jetpack Compose that solves these challenges through:
Decoupled BLE Beacon Decoding: Scans and decodes BLE advertising packets on-the-fly from a wide variety of sensor beacons (including SHT40 temperature/humidity, Lux, dissolved oxygen/water quality, ammonia, step counter, and accelerometers) without requiring active pairing.
On-Device Machine Learning: Integrates a local TensorFlow Lite machine learning model that takes raw 18-channel optical spectrometer reflectance data, interpolates it into 1,201 data points (from 400 nm to 1000 nm), and instantly predicts Nitrogen, Phosphorus, Potassium (NPK), and Organic Carbon levels directly on the edge.
Dual-Joystick Remote Control: Features a landscape gaming dashboard that connects to robots via Classic Bluetooth, steering them using interactive joysticks while showing live sensor readings.
Gamified Proximity Learning: Implements games like "Hunt the Heroes" that utilize Bluetooth signal strength (RSSI) to guide users physically to hidden beacons.
Offline-First Data Sync: Stores logs locally and leverages Firebase Firestore alongside Android WorkManager to reliably sync sensor readings to the cloud when internet connectivity is available.
Challenge :
The primary technical hurdle was implementing real-time, on-device scientific interpolation and ML inference. Raw spectral data from the optical sensor only provides 18 sparse readings, which had to be mathematically mapped to 1,201 wavelengths in milliseconds to run through the TensorFlow Lite model without freezing the UI thread. Additionally, we had to ensure thread safety while managing three simultaneous operations: high-frequency BLE scanning, continuous Bluetooth Classic socket writes for low-latency joystick control, and background database synchronization during intermittent network access.
Summary :
BLE Sense is a robust, edge-computing IoT mobile application that bridges the gap between hardware sensors and smart analysis. By combining BLE advertisement decoding, classic Bluetooth remote control, and localized TensorFlow Lite machine learning, the app provides real-time, offline-first soil health metrics, interactive device controls, and engaging educational games in a clean, modern user interface.
More Projects
App Development
BLE Sense (Android App)
BLE Sense is a comprehensive application that leverages Bluetooth Low Energy (BLE) technology to enable low-power, efficient communication between Android devices and BLE peripherals. Designed to meet the evolving needs of modern IoT ecosystems, BLE Sense excels at real-time data exchange with minimal latency.
Year :
2024-25
Industry :
Sensors
Client :
Annam.ai, IIT Ropar
Project Duration :
15 weeks
Problem :
Interfacing with multiple distinct IoT sensors typically requires separate proprietary applications and constant manual device pairing, which is slow and consumes significant battery. Furthermore, checking soil nutrient levels (NPK and Organic Carbon) traditionally relies on expensive laboratory testing or cloud-dependent analysis, making on-the-spot agricultural diagnostics impossible. Additionally, DIY robot controllers often lack real-time telemetry display, and there is a general lack of interactive, gamified ways to teach students how Bluetooth signal propagation and proximity sensing work in the real world.
Solution :
We developed BLE Sense, a unified, native Android application built with Jetpack Compose that solves these challenges through:
Decoupled BLE Beacon Decoding: Scans and decodes BLE advertising packets on-the-fly from a wide variety of sensor beacons (including SHT40 temperature/humidity, Lux, dissolved oxygen/water quality, ammonia, step counter, and accelerometers) without requiring active pairing.
On-Device Machine Learning: Integrates a local TensorFlow Lite machine learning model that takes raw 18-channel optical spectrometer reflectance data, interpolates it into 1,201 data points (from 400 nm to 1000 nm), and instantly predicts Nitrogen, Phosphorus, Potassium (NPK), and Organic Carbon levels directly on the edge.
Dual-Joystick Remote Control: Features a landscape gaming dashboard that connects to robots via Classic Bluetooth, steering them using interactive joysticks while showing live sensor readings.
Gamified Proximity Learning: Implements games like "Hunt the Heroes" that utilize Bluetooth signal strength (RSSI) to guide users physically to hidden beacons.
Offline-First Data Sync: Stores logs locally and leverages Firebase Firestore alongside Android WorkManager to reliably sync sensor readings to the cloud when internet connectivity is available.
Challenge :
The primary technical hurdle was implementing real-time, on-device scientific interpolation and ML inference. Raw spectral data from the optical sensor only provides 18 sparse readings, which had to be mathematically mapped to 1,201 wavelengths in milliseconds to run through the TensorFlow Lite model without freezing the UI thread. Additionally, we had to ensure thread safety while managing three simultaneous operations: high-frequency BLE scanning, continuous Bluetooth Classic socket writes for low-latency joystick control, and background database synchronization during intermittent network access.
Summary :
BLE Sense is a robust, edge-computing IoT mobile application that bridges the gap between hardware sensors and smart analysis. By combining BLE advertisement decoding, classic Bluetooth remote control, and localized TensorFlow Lite machine learning, the app provides real-time, offline-first soil health metrics, interactive device controls, and engaging educational games in a clean, modern user interface.
More Projects
App Development
BLE Sense (Android App)
BLE Sense is a comprehensive application that leverages Bluetooth Low Energy (BLE) technology to enable low-power, efficient communication between Android devices and BLE peripherals. Designed to meet the evolving needs of modern IoT ecosystems, BLE Sense excels at real-time data exchange with minimal latency.
Year :
2024-25
Industry :
Sensors
Client :
Annam.ai, IIT Ropar
Project Duration :
15 weeks
Problem :
Interfacing with multiple distinct IoT sensors typically requires separate proprietary applications and constant manual device pairing, which is slow and consumes significant battery. Furthermore, checking soil nutrient levels (NPK and Organic Carbon) traditionally relies on expensive laboratory testing or cloud-dependent analysis, making on-the-spot agricultural diagnostics impossible. Additionally, DIY robot controllers often lack real-time telemetry display, and there is a general lack of interactive, gamified ways to teach students how Bluetooth signal propagation and proximity sensing work in the real world.
Solution :
We developed BLE Sense, a unified, native Android application built with Jetpack Compose that solves these challenges through:
Decoupled BLE Beacon Decoding: Scans and decodes BLE advertising packets on-the-fly from a wide variety of sensor beacons (including SHT40 temperature/humidity, Lux, dissolved oxygen/water quality, ammonia, step counter, and accelerometers) without requiring active pairing.
On-Device Machine Learning: Integrates a local TensorFlow Lite machine learning model that takes raw 18-channel optical spectrometer reflectance data, interpolates it into 1,201 data points (from 400 nm to 1000 nm), and instantly predicts Nitrogen, Phosphorus, Potassium (NPK), and Organic Carbon levels directly on the edge.
Dual-Joystick Remote Control: Features a landscape gaming dashboard that connects to robots via Classic Bluetooth, steering them using interactive joysticks while showing live sensor readings.
Gamified Proximity Learning: Implements games like "Hunt the Heroes" that utilize Bluetooth signal strength (RSSI) to guide users physically to hidden beacons.
Offline-First Data Sync: Stores logs locally and leverages Firebase Firestore alongside Android WorkManager to reliably sync sensor readings to the cloud when internet connectivity is available.
Challenge :
The primary technical hurdle was implementing real-time, on-device scientific interpolation and ML inference. Raw spectral data from the optical sensor only provides 18 sparse readings, which had to be mathematically mapped to 1,201 wavelengths in milliseconds to run through the TensorFlow Lite model without freezing the UI thread. Additionally, we had to ensure thread safety while managing three simultaneous operations: high-frequency BLE scanning, continuous Bluetooth Classic socket writes for low-latency joystick control, and background database synchronization during intermittent network access.
Summary :
BLE Sense is a robust, edge-computing IoT mobile application that bridges the gap between hardware sensors and smart analysis. By combining BLE advertisement decoding, classic Bluetooth remote control, and localized TensorFlow Lite machine learning, the app provides real-time, offline-first soil health metrics, interactive device controls, and engaging educational games in a clean, modern user interface.