MIT develops self-driving scooter; will drive you around malls and airports

We’ve all heard of self-driving cars and how they’ll be taking over our jobs and revolutionizing transportation in the next five to ten years, but how many people have heard self-driving scooters? Researchers at MIT have developed just such a system, alongside a self-driving golf cart, designed to help mobility-impaired people get around.

In a collaboration between MIT and Singapore, researchers created a standard, adaptive and modular autonomous driving system, that can be utilized on different types of vehicles. While starting off on cars for regular consumers, the system was then brought over and tested on smaller use cases, for very specific user needs.

For example, the self-driving scooter, was tested indoors, where the system needed to account for its position inside of the building, find its way across largely identical corridors and, of course, take care to avoid bumping into any people or furniture. The autonomous golf cart on the other hand, is seen as a reliable way to get across a parking lot or a park.

The test came with important challenges for the modular system, but also allowed it to prove some of its best assets. By utilizing the same underlying system, researchers were able to use data between devices and vehicles. For example, the scooter learned the lay of the land from grunt work that had been done earlier by a self-driving golf cart. Once the algorithm gets better, or learns something new on one vehicle, those advances can quickly be copied over to all other vehicles.

Researchers envision a near-term future where such light vehicles could be set-up in malls, airports, public institutions, or places around town, to help those with mobility issues. Interestingly enough, a survey was put out among the 100 laypeople that got to test the autonomous scooter, on how they regarded self-driving technology. After the test, trust among users seemed to go up significantly, showing that as this technology becomes more popular, there’s no reason to believe it won’t be adopted widely very quickly.

Source: MIT