projects | Lokesh Sriram

work

Markerless Mocap

This project develops a markerless motion capture system to serve as ground truth for real-time finger joint angle prediction from surface electromyography (sEMG) signals, enabling affordable hand motion analysis with applications in prosthetics and rehabilitation.I built the computer vision pipeline using MediaPipe to extract and compute individual finger joint angles from RGB camera input

High Density Electromyography (HDEMG) Sensor

This project designs and builds a 64-electrode surface array with 2mm spacing to capture motor neuron activity from muscle tissue, enabling high-resolution sEMG data collection for applications in prosthetics and neural interface research.I designed the electrode array layout and built the accompanying headstage, integrating bio-signal amplifiers to condition and read out muscle activity signals from the array.

Model Reference Adaptive Control (MRAC) for Drone

Drone motor failure poses a critical safety risk, and adaptive control strategies offer a promising path toward fault-tolerant flight. This project implements Model Reference Adaptive Control (MRAC) in Python and Webots to compare its performance against a standalone LQR controller under a simulated quadrotor motor failure scenario, with MRAC successfully stabilizing flight at up to 70% motor failure where LQR alone could not.

Tight Learned Inetrial Odometry (TLIO) Data Loader

IMU drift is a major obstacle in long-duration robot navigation, degrading localization accuracy over time. This project builds a data pipeline using a VICON motion capture system and D-Flow treadmill to collect ground truth motion profiles for the Digit robot, then leverages that data to develop a Python data loader for the TLIO framework, reducing IMU drift by 40 cm/min. As a research assistant, I built both the data collection pipeline and the loader implementation.

Automotive Control Methods

Utilizes Python and Webots to test different control methods (PID, Full State,LQR,EKF + SLAM) on a Tesla car around a set track. Currently in EKF with SLAM state

Electric Vehicle (EV) Battery Management Model

Modelling and Optimization of EV charging and discharging with Vehicle to Home (V2H) capabilities alongside solar panel, HVAC, and stationaary battery systems on a residential scale. Goal is to reduce grid usage and feedback

Power Over Ethernet (POE) Lighting System Implementation

Designed and deployed a Direct Current (DC) lighting system. Determined new system uses 20% less energy on larger scales compared to AC Lighting systems using custom built current data aquicition system based on R-Pi. (Led to my first lead author publication)

Micro Electro Mechanical System (MEMS) mic response and EQ study

Utilizes I2s Mems microphone, raspi, and frequency manipulation to create telephone effect and quantify freuqency change

Custom Data Power Acquisition System Using Hobby Electronics

Designed and deployed power-sensing system leading to $500 savings over TED Spyder - led to second author publication to ASHRAE Transactions

Phase Sensitive Detection

Applies Phase Sensitive Detection to microphone data and analyses its effectiveness

Safety Printed Circuit Board (PCB) for Direct Current (DC) distribution panel

Designed a PCB that served as a remote control switch controled by R-Pi - allowing for smart logic to turn on/off 340 V DC bus and increasing safety

TerraProbe

Designed, machined, and assembled a 12-inch soil coring robot to improve sample integrity for agricultural research

Formula Society of Automotive Engineers (FSAE) - Aerodynamics

Revolved around Computer Aided Design (CAD) of moulds and parts in Siemens NX and Computer Aided Manufacturing (CAM) in Fusion 360 for gantry, water jetting, milling, and laser cutting machines in Fusion 360 to machine molds.

Pan Heat Distribution Analysis

Conducted analysis on heat distribution of different material pan surfaces in order to determine which would be best suited (even distribution) for cooking

Flicker Noise Exploration

Determined how impedance matching and enclosures affect 1/f noise and signal quality

Warehouse Inventory Management System (WIMS)

Utilizes JAVA and MySQL to create a WIMS for client Ms.XX from an education company named Karadi Path

entrepreneurship

Clara AI

Startup that aims to improve patient workout experience and connection to physical therpists using computer vision and natural language processing

ParkVue

A Pitch for an Internet of Everything (IoT) based Parking Management System that won $10,000 in the John Martin Entrpreneurial Center (JMEC) competition.