Embedded Systems & IOT

Low Cost Static Gesture Recognition

Our current work in this area includes research on low cost , low power wireless static gesture recognition systems that builds on our earlier research work in this area.

Currently we are exploring the use of Nordic Nrf51822 series with our custom firmware to enable low power, high efficiency wireless communication enabled gesture recognition systems on low power ARM platforms.

We are also working to enhance the user experience of the gesture recognition system using accelerometers and gyroscope.

Research is also underway to develop generic light weight gesture recognition engine for multiple sign languages that can be run on embedded platforms.

Group Members: Ateendra Ramesh, Balasubramanian T, Gautham Krishna G, Ajay Kannan, Aditya Vikram Sharma, Aparokshith Rao, Dinesh Sreekanthan, Karthik Nathan, Lakshminarasimhan Srinivasan, Sidharth, Anupkrishnan, Adithya Ganesan

Earlier work in this area includes:

Smart Devices

Device Identification and Recognition

Our current work in this area includes research on non intrusive,  low cost device recognition and identification systems that can run on embedded platforms capable of device  identification in a typical low income urban and semi urban setting in developing economies like India. This work builds on our earlier research work in this area. The test bench for this system was build on the Arduino platform.

IoT Smart Assistive Living using Wearable Devices

This project aims at designing a wearable solution using accelerometers for assistive living and explore their usage in preventive healthcare or gesture based applications. After careful analysis of the dynamics and gestural attributes of human gesturing, we have decided on using Dynamic Time Warping (DTW) to extract gesture templates from the accelerometer readings. Currently we are researching on low foot-print implementations of DTW on embedded platforms. Studies are also being done on deciding optimum sampling rate for accelerometers and quantisation techniques to accurately interpret the readings.

The proposed system is to be implemented as part of a body area network where the individual sensors would communicate with a master through BLE to enable battery friendly computations and operations. The sensors would be places on various locations on the body, to sense different parameters, the consortium of which would be used for intelligent actuations. The MSP430 platform which possess sleep state capabilities is being explored for use for low power implementation.

Group Members: Aswin Natesh, Aashish Kumar Jain S, Gosakan Srinivasan, Prahalathan Sundaramurthy, Raghavendran Thiruvengadam, Rohan Ganpati, Syed Shahbaaz

Earlier work in this area includes:

Smart Energy

Microgrid – Agent modeling, MAS systems, Solar PV – Hybrids and Clusters

Our current work in this area includes research on MAS modeling for microgrids in a rural Indian environment. Further due to the unique nature of such systems a number of agents need to be remodeled with region and source specific characteristics. Load servicing and optimizing, role of storage in such systems and intelligent dispatch of energy are other areas that are being actively researched.

Group Members: Gandhi Rajan, Meghana, Rohan Ganpati, Roshith E, Sai Shankar M, Sivaroophini, Vetrivel

Earlier work in this area  includes: