A Smarter Way For Autonomous Vehicles To Plan Their Route

The traveling salesman problem has been a staple of computational logic for many years, but with the rise of GPS applications, the challenge of finding the fastest route has come on leaps and bounds.  Nonetheless, problems persist, and a new paper reminds us that for autonomous vehicles to thrive, passengers have to have faith that they will navigate effectively.

The perfect route is still incredibly difficult to ascertain, but the paper shows the progress that’s being made, with the researchers devising a new method of optimizing the tracking of self-driving cars that reduces the errors while maintaining a low computational load.

The researchers highlight how human drivers process large quantities of information as they drive, including speed, expectations of any route, safety factors, and so on, with all of this going into determining the best route to take.  Sometimes speed will be the priority, other times comfort, and this has been difficult to replicate in autonomous systems.

Tracking progress

A major aspect of this problem is monitoring the vehicle to ensure it follows the desired route as closely as possible in an appropriate timescale.  This is the kind of tasks that humans perform routinely, but the maths behind it is incredibly complex to encode into a machine.  Solutions that have been proposed to date typically require vast computing power.

“With an autonomous vehicle, all this has to be performed in what we’d call the ‘brain’ of the autonomous vehicle,” the researchers say. “We set ourselves a challenge that is simple to state but hard to achieve with respect to trajectory planning: A passenger in a self-driving car has to feel as if it were driven by a human.”

The team believe their new method is able to reduce the computational requirements while maintaining accurate tracking, including the position, velocity and acceleration of the vehicle.

The next step is to try and make the system more widely available, whilst also incorporating a wider range of variables, such as tire forces and side slipping that are common in harsh road conditions.

Facebooktwitterredditpinterestlinkedinmail