Apoorva Suresh

“Deep Learning Based Sensor Fusion for 3D Obstacle Detection in Autonomous Driving” • Analysis of state-of-the-art deep learning approaches for 2D / 3D object recognition and depth estimation (mask RCNN and point pillars) • Development of architectures for neural networks, as well as their training, data preparation and algorithm evaluation • Runtime optimization with TensorRT on a High Performance Computers and on NVIDIA Jetson AGX Xavier board • Tools: Python, OpenCV, Keras, Tensorflow, TensorRT

"Machine learning and Algorithm development for Autonomous driving” • Development of a camera-based deep-learning algorithm for object detection • Development of a Python module for estimating the size and position of obstacles based on camera detection and the depth of a lidar sensor using classical computer vision and sensor fusion methods • Tools: Python, Keras, Tensorflow, OpenCV

• Incorporation of geometric and logical road description into the OpenDrive specification and implementation of an XML, Matlab interface • Tools: OpenDRIVE, Matlab, XML

ADAS, Autonomous driving, Machine Learning Master Thesis at ZF Friedrichshafen AG: “Deep Learning Based Sensor Fusion for 3D Obstacle Detection in Autonomous Driving” Average grade point: 1,2

Average grade point: 1,4