In spite of all the intelligent sensors and the technology that incorporate the
autonomous carsYes, there is still something missing and that a newly licensed driver already has: social awareness. The machines still
they see the rest of the drivers as a set of data They have to analyze and not as human beings who have a specific motivation and personality. A group of researchers from the MIT Laboratory of Computer Science and Artificial Intelligence has been exploring whether autonomous cars can be
program to classify the personalities of other drivers, which will help them to better predict how they will behave in different situations and, therefore, drive more safely.
Initially, the researchers used tools of social psychology to classify driving behavior with respect to how selfish or selfless a particular driver is, as explained by MIT Technology Review. Then, the system estimates these characteristics of drivers – based on their behavior – to create real-time driving paths for autonomous cars. By testing their algorithm, the team demonstrated that they can better predict the behavior of other cars by 25%. For example, when doing simulations in which an autonomous car should turn left, the car knew what to expect when the approaching car had a more egosta driver and that it could turn when the other car was more altruistic.
The system was trained to try to understand better when it is appropriate to exhibit different behaviors. "Even the most respectful human drivers know that certain types of actions, such as making a lane change in a heavy traffic jam, require a time to be more assertive and decisive," the publication reads.
"Although the results are not yet robust enough to be implemented on real roads, the system could have some interesting use cases," and not only for cars that drive alone, but also for those with intelligent driving assistance. Thus, the car can tell you what kind of drivers are in other vehicles that are passing near you.
Discover intentions to understand behaviors
"Working with and around humans means discovering their intentions to better understand their behavior," Wilko Schwarting, lead author of the study, explains to the publication. "People's tendency to be collaborative or competitive often extends to how they behave as drivers. Our goal was to quantify it."
A central problem with current autonomous cars is that they are programmed to assume that all humans act in the same way. This means that, among other things, they are quite conservative in their decision making. While this precaution reduces the possibility of fatal accidents, it also creates bottlenecks that can be frustrating for other drivers. "Creating more humane behavior in autonomous vehicles is critical to the safety of passengers and surrounding vehicles, since behaving in a predictable manner allows humans to properly understand and respond to the actions of intelligent vehicles," says Schwarting.
For the next phase of the investigation, the team plans to work to apply its model to pedestrians, bicycles and other agents in driving environments. In addition, they will investigate other robotic systems that act with humans, such as domestic robots, and integrate social variables into their prediction and decision-making algorithms.