Abstract: Autonomous driving (AD) is a challenging and fascinating area of research with the established goal of shaping the future of transportation. The AD software necessary to accomplish this objective often requires a complex solution that integrates multiple components inside the vehicle, such as a perception component (how the vehicle perceives the world through its sensors), a fusion component (how this information is combined), or a planning component (which behavior the vehicle will choose according to the data). While ML takes intuitively a center role in shaping the technology that is deployed in all these components inside the vehicle, its role outside of the vehicle and within the scope of Data Science is less obvious, yet extremely important. In this talk we will illustrate some of the projects and challenges faced by the Data Science team behind the autonomous project of Robert Bosch and Daimler. Three main projects will be discussed. One project describes the creation and running operation of a “data center”, which provides on-demand and routine data analyses to developers and management. Second, we will present a number of solutions we implemented to strengthen root cause analysis and perform anomaly detection. Third, we will discuss some of the algorithms and the applications of scenario identification using data collected from our autonomous vehicles. All these projects will be connected to the recent testing operation Daimler and Bosch run in California where a first version of their AD kit was deployed via a fleet of S-class Mercedes.
Bio: Stefano Cosentino (PhD) is Senior Data Scientist at Bosch, where he works within the Autonomous Driving team in collaboration with the Mercedes Benz Research Center. Born and raised in Italy, he spent most of the past decade between London and Cambridge, England. In the UK, he studied and worked in the academia researching signal processing, applied machine learning, and psychophysics. Recently, he moved to the San Francisco Bay Area to join the Bosch research center in North America. At Bosch, Stefano leads the Data Science team and works on applied machine learning topics related to scenario identification, root cause analysis, anomaly detection, and data analysis.