Academic Internship - Sim2real transfer of DeepLearning models trained on a driving simulator

<h3>Company Description</h3><p>At Bosch, we shape the future by inventing high-quality technologies and services that spark enthusiasm and enrich people’s lives. Our promise to our associates is rock-solid: we grow together, we enjoy our work, and we inspire each other. Join in and feel the difference.</p><p>The company' success lies in its highly specialized and innovative team, and on the technological know-how which makes Bosch the leading supplier in the automotive market.</p><p>Bosch is becoming one of the world's leading companies in the field of the Internet of Things (IoT). At the Bosch Research &amp; Development Center in Braga, more than 350 engineers develop car multimedia solutions and the most innovative localization, perception and safety sensors that are helping to shape connected and autonomous mobility. From hardware to software to physics, we have all kinds of opportunities waiting for you.</p><h3>Job Description</h3><p>Autonomous driving Deep Learning perception algorithms benefit from accurate 3d point cloud data/labels. Large amounts of real driving data is expensive to obtain, especially for corner-case situations.</p><p>Manual, accurate labeling of 3D data (e.g. 3d bounding boxes, point cloud semantic segmentation, lane-estimation, etc...) is expensive and time consuming to obtain. Auto-labelling techniques try to provide higher quantity of data, but still lag in terms of precision.</p><p><strong>Your contribution to something big:</strong></p><ul><li>Setup a framework to generate labeled data from simulated scenes with parametrized properties (e.g. day/night, weather conditions, sensor properties/noise, corner situations...).</li><li>Compare performance of DL models trained on simulated data, with ones trained exclusively on real data. </li><li>Investigate performance drop and approaches mitigate to sim2real gap (e.g. fine-tuning on real data, synthetic data augmentation...).</li><li>Investigate scaling of performance on real data wrt. to amount/percentage of simulated data.</li></ul><h3>Qualifications</h3><p><strong>What distinguishes you:</strong> </p><ul><li>Enrolled in a Masters program in Computer Science, Robotics, Physics, or a similar technical field of study;</li><li>Solid C++ and/or Python programming skills;</li><li>Experience in at least one of the following domains: machine learning, robotics, computer vision, LiDAR, Radar;</li><li>Familiarity with deep learning frameworks (e.g. Tensorflow/PyTorch) is a plus;</li><li>Good level of English (spoken and written);</li><li>Sense of Team Work.</li></ul><h3>Additional Information</h3><p><strong>What we offer you: </strong></p><p>Challenges in an international environment | career opportunities in one of the best companies in the world | flexible hours | medical services | discounts for employees | sports and health related activities | canteen | good access to public transports | space for creativity.</p><p><strong>Success stories don´t just happen. They are made... </strong></p><p><strong>Make it happen. We are looking forward to your application!</strong></p>