Tejas Hadkar

Concepts, Design and Software Development Software implementation of derived concepts in autonomous driving using C++, Wrappers for third-party libraries, Parallel Programming using CUDA, Project Management Tools: Confluence, Jira, Bitbucket, Bamboo Source Code Management tools: Git, Winmerge Unit testing using gtest C++ framework

Segmentation of 3D point clouds from a LiDAR using deep neural networks 1. Simulation of Velodyne LiDAR sensor using Unity game engine 2. Generate dataset using developed LiDAR simulator 3. Analyze, Clean and prepare dataset 4. Development of Deep Neural Network model and train for multiple object detection and localization 5. Test and Evaluate developed model

Prediction of Lane Change Maneuver of driver using head movements and glances Responsible for : - Select features from collected measurements - Cleanse, order and reshape data - annotate and pre-process data for trainng - build Recurrent Neural Network model - train it using developed dataset with the help of GPU - Analyze and evaluate results

Responsible for analyzing response and performance of newly fabricated raw humidity sensors. Measurement automation is achieved by controlling LCR meter from NI LabVIEW. For calibration purpose an industrial sensor unit is interfaced with Arduino over I2C bus and measurements are captured in csv files from LabVIEW connected to Arduino over GPIB bus. Measurements were studied and analyzed for fast response time, good range of electrical signal and reproducibility. ISBN 978-3-00-054047-9

- Image Processing and Pattern Recognition - Design of Heterogeneous Systems - Real Time Systems - Design of Software for Embedded Systems - Software Platforms for Automotive Systems - EDA Tools - Hardware-Software Co-design - Automotive Sensor Systems

Microprocessors, Micro-controllers, Digital Electronics, Microelectronics, Computer Organisation and Architectures