New AI system makes autonomous cars smarter and faster

The problem: Existing technology is not fast or accurate enough

Safe driving isn’t just about what you’re looking at; It’s about predicting what might happen next.

Being able to quickly and accurately estimate the future movements of cars and people around them is crucial for self-driving cars.

Even a small mistake or delay could lead to a terrible accident. Currently, the technology in these cars is not as fast and accurate as it needs to be, especially in heavy traffic.

QCNet: The fast and intelligent solution

A team of researchers from the City University of Hong Kong led by Professor Wang Jianping has made a major breakthrough.

They have developed a new AI system called QCNet that is much faster and more accurate than anything previously available. QCNet uses so-called “relative space-time” principles to be more efficient.

This basically means that the AI ​​doesn’t have to repeat a lot of calculations every time something changes in its view. This allows it to make real-time predictions about what cars and people around it will do next, all without slowing down.

QCNet also understands how cars and people interact on the road, making its predictions more reliable. The team used two sets of real-world driving data from US cities to test how well QCNet does its job.

The results were impressive: QCNet was able to predict the movement of road users up to six seconds into the future and was faster and more accurate than 333 other methods.

What’s next: Making self-driving cars even safer

What does all this mean for the future of self-driving cars? Well, QCNet will be part of the technology used in autonomous driving systems to make them even safer.

It is also planned to use this technology for traffic simulations and other safety functions.

Professor Wang says this new AI system will help self-driving cars better understand their surroundings, make more human-like decisions and be much safer overall.

The results of this research have not only impressed the academic world, but will also find application in real systems.

Hon Hai Technology Group (aka Foxconn) and Carnegie Mellon University in the US have collaborated on this research and plan to incorporate QCNet into their self-driving technology to improve safety and efficiency.

In a world that is becoming increasingly dense, intelligent and fast solutions such as QCNet are becoming increasingly important.

It’s a big step forward to ensure that the self-driving cars of the future are as safe as possible for all road users.

follow us on Twitter for more articles on this topic.

Source: City University of Hong Kong