When it Comes to AI in Autonomous Vehicles, it Better be Perfect
Although Level 5 Autonomy for autonomous vehicles (AVs) is not widespread yet, the technology is being actively deployed in next-generation cars today and will lay the foundation for the autonomy of tomorrow. The technology being used in testing AVs today is extremely impressive — the field has figured out around 80% of the autonomy puzzle. However, the last 20% is going to be the most difficult and take much longer than the pace of development for today’s tech, because safety is the number one priority.
One of the major hurdles on the road to full autonomy is the ability to program AI systems to react to real situations in real-time. Current AVs are going through a two-part preparation process of training and inferencing — both of which are done in controlled environments or simulations — but there are only so many scenarios that can be imagined and tested before the real world throws a curveball at these cars. Giving cars the ability to read and react to these situations, then share that information with other AVs, will be key to achieving full autonomy. Even though the challenges ahead are great, the benefits can usher in a powerful, life-changing technology.
To learn more about this topic please follow the link for the original source: