Chess StetsonChess completed his doctoral and post-doctoral work in computation and neural systems at Caltech, and previously earned an AB in physics from Harvard. He has more than a dozen peer-reviewed publications in experimental and computational neuroscience, three patents issued and three pending, and has spent years building and refining dRISK’s core knowledge graph IP and providing AI/ML solutions to Fortune 500 companies, large integrated healthcare providers, financial services firms and autonomous vehicle companies. He also occasionally appears as a neuroscience expert on National Geographic’s Brain Games.
Risk-aware autonomous vehicle perception
How do we make a safe, fully driverless car? While AVs tend to perform adequately during normal driving conditions, they still fail on edge cases – the many hard, high-risk scenarios that are individually unlikely but collectively make up all the risk. In this presentation, we introduce a novel solution to training on edge cases, all the way to the level of the perception system, resulting in a 6x improvement in detection of high-risk events with 2x improvement in detection confidence. This makes AVs safer and thus commercially viable.