Improving Autonomous Vehicle Decision-Making for Safety and Efficiency

Partnerships between businesses must use 5G networks' high throughput and low latency to improve the speed and reliability of shared perception for autonomous vehicles.

FREMONT, CA: Artificial intelligence in self-driving vehicle technology makes it more reliable and effective. Vehicles could get new information and make decisions much faster, leveraging potential systems. When a self-driving car lacks enough information and doesn't know whether to brake or turn, handing over control might take a while. At that point, it's too late to do anything. Things change quickly in cities, so three seconds is also a long time. The standard change between three and ten seconds in car self-driving is vast and scary, so when the system realizes it cannot drive the car anymore, it is called a handover situation.

A big problem for the safe commercialization of autonomous vehicles is that they can't quickly see and understand what's happening around them. It is because computers can't do as much, and the environment isn't always predictable. It will let them set the processing bar (which includes making decisions) at about 20 milliseconds so that vehicles can update their data and make faster, safer decisions in real time. There is a need to make a big difference by limiting the information a vehicle receives to what is most important.

Various automobile system works by combining onboard sensors, which read the vehicle's immediate surroundings. It helps to track moving objects, with a camera and sensors mounted on nearby fixed things, like lampposts and buildings at intersections, to give a much broader view. There might be a limited field of view if, for example, a pedestrian walks in front of a bus or a cyclist comes out of nowhere. For instance, if a kid chases a ball across a street, a sensor could pick up on that movement up to 100 meters away before the car gets there.

The drivers take action ahead of time, like stopping or turning, instead of waiting until they see a moving object and understand. The use of artificial intelligence to filter information is one of his most important contributions. Instead of considering everything in a vehicle's field of view, like a human driver, the system's algorithm only looks at important information, like moving objects in its path and features like traffic lights and signs. The cooperative sensor systems can work in different urban environments with higher odds of occlusion or moving objects.

Companies need to get into partnerships to use the low latency and high speed of 5G networks to make shared perception for self-driving cars faster and safer. The fast-growing technology can get used in warehouses, where robots could move around close to people or, with their help, for mobile robots connected to a network and do maintenance and repairs in changing environments.