1. Research and Development of low cost quadruped robot for mine inspection:
● Developed and tested various quadruped robot's sensors and actuators hardware interface programs using ROS, Python, and C++.
● Implemented and tested robot localization algorithms (based on Extended Kalman Filter) to estimate the position of the robot in a global frame, utilizing data from an IMU, leg odometry, and a depth camera.
● RTAB-MAP and RGD-SLAM ros packages are used for visual odometry from intel realsense depth camera data.
● Developed and tested PID, MPC, and Impedance controllers to manage the dynamic locomotion of a quadruped across various terrains.
● Researched and developed hierarchical optimization control to track the desired motion of the base and the desired motion of the robotic manipulator mounted on the robot.
2. Research and Development of EMG based Robotic Brace for stroke patient rehabilitation and assistance:
● Responsible for complete software development and leading a cross functional team of 4 engineers for this project.
● Developed C++ programs to interface multiple EMG (Electromyography) sensors , motors and its control.
● Implemented Task space Impedance control to control the arm using TSID and Pinnochio libraries.
● Implemented MPC control in C++ and tested it on robot brace for the trajectory tracking of the arm.
● Implemented signal processing algorithms to filter multiple EMG sensor data.
● Developed AI/ML multiclass classification models to infer user bicep and tricep activations using libraries like Tensorflow and ONNX runtime.