While tech companies battle to be the first to create fully autonomous vehicles, there is a quiet evolution happening in the background that is building the foundation for these vehicles: data collection, with insights powered by artificial intelligence and machine learning. For autonomous vehicles to reach the market and remain successful, a perfectly personalized experience for self-driving car users is essential. Using AI to learn driver habits and personal preferences and collecting a variety of data gives tech companies powerful business intelligence, that will help them shape successful self-driving technologies and make way for fully autonomous transport.John Cordell, Chief Product Officer at Xevo AI, says "self-driving cars may not be ready to hit the road just yet, but the technology of the smart cars of tomorrow can be implemented in cars today." John will be speaking at the Machine Intelligence in Autonomous Vehicles Summit in San Francisco next week, sharing expertise on the current technology of connected cars and how that will impact the future of autonomous vehicles. I caught up with him ahead of the summit to learn more about his work.What started your work in autonomous vehicles?I founded an AI startup, Surround.IO in 2014 focused on the general problem extracting meaning from video in real-time using low-cost devices at the edge and cloud computing for aggregation and model creation. Connected vehicles were a natural application for the core technology we had developed. What are the key factors that have enabled recent advancements in autonomous vehicles?Three main components: sensor technology, the advancement in machine learning, and Moore’s law. In particular, the rediscovery of back propagation in the mid-80’s coupled with modern GPUs created the ability to train large networks that are capable of doing significant and useful acts of classification on complex input signals.
How important is data collection and business intelligence for the continued progression of autonomous vehicles?
Data collection is crucial to the development of successful autonomous vehicles. Without it, viable autonomous vehicles would not exist. The reason for that is the fundamental inversion that machine learning has created between the value of data vs. the value of algorithms. It is now the case that developing the best and most useful software demands large data sets. Autonomous vehicles must have software that can handle the real-world, which implies ML techniques, which in turn implies massive amounts of data.
How will current connected vehicle technology shape self-driving cars of the future?
Probably in ways we can’t predict. But one thing is clear: autonomous solutions will greatly benefit by having tens of millions of connected vehicles cooperating to achieve a safe and enjoyable autonomous vehicle experience for passengers.
What developments can we expect to see in autonomous vehicles in the next 5 years?
Two words: bucket seats. Plus sophisticated contextual voice assistance, adaptive driving styles based on fine-grain situational context and user preferences, and perhaps novel ownership models created by large scale driverless ride-sharing vehicles.
John Cordell will be speaking at the Machine Intelligence in Autonomous Vehicles Summit, taking place alongside the Machine Intelligence Summit, in San Francisco on 23-24 March. Meet with and learn from leading experts in autonomous vehicles, IoT, the connected car, machine learning methods and predictive analytics. Register to attend here.
Other confirmed speakers include Gaurav Bansal, Senior Researcher at Toyota InfoTechnology Center; Guan Wang, Machine Learning Engineer at NIO; Luca Rigazio, Director of Engineering at Panasonic SV Lab; and Pratik Brahma, Machine Learning R&D at Audi/VW. View more speakers and topics here.
The Machine Intelligence in Autonomous Vehicles Summit will also take place in Amsterdam on 28-29 June. Book your place here.