Abstract: To safely navigate roads and adapt to our ever-evolving cityscape, self-driving vehicles need a deep understanding of the world around us. We’ll explore how Waymo uses real-world and simulated data with deep learning to unlock new capabilities and build safer autonomous vehicles. We’ll also cover how Waymo uses data to develop machine learning at scale as it expands to new cities and geographies.
As the only autonomous vehicle company that designs the full suite of self-driving hardware and software in-house, Waymo has developed an integrated system that allows each Waymo vehicle to process a large and diverse set of sensor data, ultimately allowing our vehicles to make informed, real-time decisions. We’ll touch on the benefits of designing and developing our hardware (LiDAR, radar, and cameras), software and manufacturing systems, how deep learning is used in various parts of self-driving from mapping, to perception, behavior prediction, and more, and what machine learning infrastructure is required to develop self-driving technology at scale.
Bio: Sacha Arnoud is a senior director of engineering at Waymo, a self-driving technology company with a mission to make it safe and easy for people and things to move around. In his role, Sacha oversees the company’s perception efforts from multi-sensor configuration to signal processing, advanced and industrialized machine learning to complex scene semantics understanding. Sacha has twenty years of industry experience in Silicon Valley, from leading development on Sun Microsystems first distributed large-scale archival storage system, to pioneering the use of Deep Learning-based computer vision at Google Maps using Street View imagery. Sacha earned a bachelor’s degree from the Ecole Polytechnique, France in 1998 and a master’s in telecommunication systems and computer science from Telecom ParisTech, France in 2000.