A new system classifies behavior to help self-driving cars better anticipate what other cars will do so that they can drive more safely.
Our autonomous future: cars without drivers, drones and robot vehicles will bring unprecedented change and affect every industry
Drivers without cars are the world’s most prominent example of AI and robotics in action. But trucks, semi-trucks, vans and drones without drivers are ready to unleash a new level of automation in the company.
A team of researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has developed a system to determine whether self-driving cars can be programmed to predict the driving personalities of drivers in other vehicles.
More about artificial intelligence
The system classifies the behavior of drivers to help self-driving cars better anticipate what other cars will do so that they can drive safer among them.
The researchers used an existing framework that is used for personalities known as ‘social value orientation’, which reflects the degree to which someone is selfish (‘selfish’) as opposed to altruistic or cooperative (‘prosocial’). The system then maps real-time lanes for driverless vehicles based on that measurement.
SEE: Guide of IT leaders for the future of autonomous vehicles (TechRepublic Premium)
At the moment, self-driving cars are generally programmed to assume that all people act in the same way, said Wilko Schwarting, an MIT study and lead author of the study.
For example, there may be a challenging merge scenario on a short driveway where an autonomous vehicle has to negotiate with another driver about whether and how it can perform the merge.
Schwarting often said that human drivers are either slowing down to widen the gap to allow one car to merge successfully – or speeding up to indicate that it is not OK to merge.
“An autonomous vehicle must recognize these subtle social signals of egoism or cooperation – and failing to do this not only reduces the overall flow of the traffic network but also affects the safety of the cars in that traffic,” he explained. “We wanted to create a system that makes more human-like driving possible (autonomous vehicles), by better understanding the social behavior of drivers around them.”
The researchers designed and tested an algorithm in this type of merge scenario, as well as an algorithm that makes unprotected left-hand turns. They showed that they could predict the behavior of other cars by a factor of 25% better.
One of the challenges the researchers discovered in the first test phase was that modeling human drivers is difficult, Schwarting said. “We must take into account how our own actions will influence the actions of the drivers around us.”
The SVO is a good measure of estimating the behavior of human drivers during these mergers and left-turning interactions, he said.
“It also allows us to decide how selfless (or selfish) an AV should depend on the scenario. Acting too conservatively is not always the safest option, as it can cause confusion in human drivers.”
No timeline for the implementation of the Schwarting SVO system has been said. “As a next step, we hope to try applying the model to pedestrians, bicycles, and other types of agents who are part of these environments,” he said.
“We also want to look at other robot systems that need to communicate with us, such as domestic robots that can benefit from such a system,” as well as care robots and tour guides for robots.
“The ultimate goal is to develop AVs that can more easily communicate with human drivers in real environments,” he said. “Creating more human-like behavior for them is fundamental to the safety of passengers and surrounding vehicles, because people can understand in a predictable way and respond appropriately to the actions of the GTC.”
At the moment, all the elements involved in driving are too complex for a robotic system to handle alone, according to a separate MIT study from August.
However, the authors said that the fact that people should play an integral role in the self-driving process is the current challenge because of the underlying uncertainty of human behavior as represented by any type of social interaction and conflict resolution between vehicles, pedestrians and cyclists. ”
Technical news that you can use Newsletter
We deliver the best corporate news stories about the companies, people and products that are revolutionizing the planet.
young woman using a smart phone in an autonomous car. car without driver. self-driving vehicle. heads-up display. automotive technology.
Getty Images / iStockphoto