Yash Shah

State estimation and localization expert for the Level-4 autonomous car. Developing probabilistic models based on Kalman filter, particle filter, and pose graph optimization like to deal with noisy data from GPS, IMU, and vehicle odometry and camera and Lidar visual odometry. Responsible for the sensor fusion of the multiple LiDARs, RADARs, and cameras. The work includes all three modules of temporal calibration, extrinsic calibration, and data fusion from the sensors. Tools -: ROS2, C++, Python

Markerless 6-DOF pose estimation and tracking of a moving object using an RGB-D camera and KUKA lightweight robot for a pick and place real-time application. Calibration between robot and camera, and solving forward and inverse kinematics were also part of the tasks. Localization and mapping of multiple Neobotix AGVs in a large-scale environment using SLAM. In the end, the Kuka robot was placed on top of the AGV to develop a multi-cobot doing various kinds of applications. Tools-: ROS, C++, Java, Python

Autonomous patient motion compensation while insertion of the needle, using a small-scale time-of-flight camera and UR3 lightweight robot. Built complex imaging algorithms based on feature extraction, optical flow, and ICP registration for markerless estimation of patient movements. Tools: C++, Python, MATLAB, MoveIt, OpenCV, PCL, NanoFlann, ROS.

An application on medical robotics for autonomous needle insertion inside a human skeleton using a 7-Dof lightweight robot. Tasks: RGB-D image data processing, 3-D model recording, and registration of skeleton and the surroundings, solving the kinematics of the robot, eye-in-hand calibration, trajectory planning with obstacle avoidance. Tools: ROS, Python, and C++, OpenCV, Point Cloud Library (PCL) & Open3D.