The public must be involved in self-driving vehicle deployment and adoption decisions. If self-driving technology is not controlled to safeguard public safety, there is a genuine risk of losing public faith.
FREMONT, CA: A human driver will not always be involved in the driving task at higher levels of automation. The AI self-driving program would basically replace the driver. While the technologies required to enable increased levels of automation are fast evolving, creating a vehicle that can safely and legally complete a journey without human intervention remains a significant issue. The following major obstacles must be overcome before they can be securely released to the market: technology and public acceptance.
Self-Driving Software and Machine Learning
Highly automated vehicles have self-driving software as a primary differentiator. The program is built on deep learning neural networks with millions of virtual neurons that replicate the human brain and machine learning techniques.
There is no explicit ‘if X happens, then do Y’ programming in the neural nets. Instead, they are taught to recognize and classify items from millions of videos and photographs captured in real-world driving situations. The more diverse and representative the data, the more adept they can identify and respond to various circumstances. Neural net training is similar to holding a child’s hand while crossing the street and encouraging them to learn by constant experience, replication, and patience.
While these algorithms are pretty good at detecting and classifying objects, they cannot quite match the intricacies of driving. Autonomous vehicles must detect and recognize persons and other objects and interact with, comprehend, and react to their behavior. They must also know what to do in unusual situations. The challenge of controlling the unexpected will be rather difficult to deep learning and training without many examples for all possible driving circumstances.
The public must be involved in self-driving vehicle deployment and adoption decisions. If self-driving technology is not controlled to safeguard public safety, there is a genuine risk of losing public faith. The lack of trust will harm both people who wish to utilize the technology and those who share the road with them.