The researchers claims that one hour of driving data was utilised for machine learning purposes, 92% of the vehicles arrived safely at their destinations devoid of any damage
A Boston University researcher, Eshed Ohn-Bar is now trying hard to introduce innovative ways of self-driving cars to learn and identify the technologies of safe driving, which is by observing the methods of numerous drivers on the road and speculating their response.
Self-driving cars that are backed by machine learning algorithms, a huge volume of driving data is needed for them to operate safely on the road. But, according to the researcher, if these cars could learn to drive the same way as babies do by watching others, they would need a lesser amount of intricate driving data. Of late, the researchers at the university showcased their research at the 2021 Conference on Computer Vision and Pattern Recognition in an effort to craft road safety more achievable through daily lives.
The researchers also have plans to train a large number of autonomous vehicles. They suggested machine learning algorithms by calculating the blind spots and viewpoints of numerous nearby cars in order to unleash bird's-eye view maps for autonomous vehicles. This will ultimately assist them identify various hurdles like turns without crashing, and deeply understand how chauffeurs behave while making their moves on the road.
The future autonomous vehicles could learn a lot by observing other cars on the road and the “learning by watching” machines has the potential to translate what they witness from other vehicles into their own standpoint. This implies that if both a traditional vehicle and a self-driving car are nearby, these observations will be the key to teach this algorithm how to navigate via its surroundings with full caution and safety.
