Sakshi Goyal

• Implemented a parser for extracting and translating relevant information from the OpenDRIVE format to support algorithms for decision-making and control. • Developed C++ algorithms to validate consistency between real-world sensor data and map data. • Tested and enhanced existing Python/C++ software to improve functionalities. • Implemented coordinate transformations to accommodate various projection types. • Integrated Dockerized tests into CI/CD pipeline with Jenkins

• Researched and implemented CNN and Transformer based methods to identify out-of-distribution scenarios in autonomous driving datasets, using pytorch. • Developed and trained neural networks, object detection and depth prediction • Implemented Kalman filter for 2D LiDAR/Radar sensor data, enhancing precision and accuracy. • Executed perception sensing and autonomous path planning using Carla simulator and ROS. • Implemented Named Entity Recognition using NLP models tailored to a custom dataset.

• Automated training and evaluation processes for pose estimation. • Created a visualization tool for model comparison. • Calibrated stereo and IP cameras using OpenCV. • Improved tracking speed through GPU-based optical flow estimation. • Implemented pose outlier detection and model selection using scikit-learn.

Majored in Computer Vision