Stanford team build robots able to navigate the streets

jackrabbotWhilst drone based delivery systems have gained a lot of publicity in recent times, there have also been a couple of interesting ground-based delivery platforms coming on stream that use the same automated approach.

These devices are designed to navigate to the destination via often busy footpaths, so being able to maneuver around pedestrians is going to be crucial to their success.

Meet JackRabbot

Successfully navigating such environments is particularly tricky because first, robots need to understand the way humans move.  Our conventions and habits if you like.  It’s an outcome that Stanford researchers are hoping to expedite via their latest robot, which they’re called JackRabbot.

The researchers have developed a computer-vision algorithm that is capable of predicting how people behave on a busy pavement.  The robot comes complete with an algorithm that has been trained using machine learning based upon a large number of videos showing people moving around busy streets and other public spaces.

When they tested the performance, it emerged that the algorithm out-performed a number of existing methods of predicting how humans will move.

The robot comes complete with a number of cameras and sensors and is being tested in a range of environments to try and replicate a real world use-case.

Building on driverless cars

The researchers are building upon the kind of technologies developed for driverless technology, and there is undoubtedly a need for cars to build a better understanding of human movements and behaviors.  With delivery robots and the like on the rise however, spreading this understanding out into a wider range of robots would be hugely valuable.

This could be relatively top level understandings such as how humans move on the roads, pavements and in public spaces, all the way down into the minutiae of our habits, such as chivalrous acts or how humans navigate parking lots.

Starship Technologies, the company that’s pioneered land-based automated deliveries, have highlighted this as their most prickly challenge, both in understanding how humans behave, and in humans having confidence in the robots such that they are an accepted presence on our streets.

It’s part of a growing attempt to make robotic movements more nuanced and human like, and it will be key to the success of them on both the roads and pavements.

Related

Facebooktwitterredditpinterestlinkedinmail

Leave a Reply

Your email address will not be published. Required fields are marked *

Captcha loading...