MM Blog: Autonomous Cars That Avoid Hazardous Conditions

Proponents of self-driving technology believe that autonomous cars could one day effectively eliminate most accidents on the road. But as regulators consider how to regulate the rapidly emerging technology, critics warn that current autonomous systems still have difficulties with several real-world driving scenarios, from poorly defined lane lines to lousy weather. Researchers from Georgia Tech, however, could have a solution to the latter problem.

They developed a method called model predictive path integral control, which uses onboard sensors and computers, along with advanced algorithms, to calculate the most stable possible trajectory and execute optimized handling decisions in real time — even at the edge of a vehicle's handling limits. Researchers said that the system technology essentially infuses self-driving systems with some of the expert techniques of human drivers. Georgia Tech engineers successfully tested autonomous rally cars — at one-fifth the scale of a normal car — by racing, sliding and jumping it around a dirt test track at the equivalent of 90 miles an hour.

SO, WHAT DO YOU THINK?

Could this system satisfy those that remain wary of self-driving vehicles? Could other transportation or manufacturing systems benefit from similarly fast, complex computations?

Email us or leave your comments below.