At the Massachusetts Institute of Technology developed a roboskuter

Yes, such a scooter cannot be called a small one (Source: MIT)

Many companies , both foreign and domestic, are now engaged in the creation of automatic vehicle control systems. Such systems can be either fully automatic - Google deals with similar developments, or partially (digital assistants of various types). By the way, Autopilot of Tesla Motors is a digital assistant. This is not an autopilot in the full sense of the term. The developers of the Massachusetts Institute of Technology (Massachusetts Institute of Technology, MIT) decided to create a robotic scooter. The project was developed in collaboration with organizations such as the National University of Singapore and the CSAIL (Artificial Intelligence Laboratory).

Earlier, the same team of researchers was studying the possibility of creating automatic golf mobiles. For them, a special software platform was developed, which with some changes and additions was used in the new project . Scooters with an automatic control system, according to the developers, can be used, for example, by those people who cannot walk for any reason independently. Such systems can be useful to both private users and companies.

Now developers are testing the functionality of the algorithms of the software platform indoors, although it all works on the streets. The work of scooters has already been tested in some areas of the MIT institute, where the system of corridors is rather complicated, which causes difficulties not only for automatic vehicle control systems, but also for people. “One of the places where we did the testing is the MIT“ endless corridor ”, navigating through which is a whole problem, since it is a long walk with a number of obstacles. You can easily get lost here. But our algorithms showed good performance in a new environment, ”said Scott Pendleton, one of the project participants. The length of the transition is not so great - only 251 meters.
')
The software platform consists of several layers. The first layer is the low-level algorithms that allow the scooter to navigate in a dynamic environment, avoiding, for example, people on the way. The second layer is the algorithms that control the direction of motion. The third layer is the algorithms that allow the system to identify its location. In addition, the developers have developed algorithms that, analyzing the situation around, map the terrain.

The MIT developers are sure that it is worth using a unified platform for controlling the movement of scooters, golf cars and city cars: systems are different. " In particular, the unification allows you to transfer data collected by one vehicle, other cars and scooters. All of these vehicles will be able to work with a single database, in contrast to situations where different vehicles work with different software platforms. Data transfer from one golf mobile to another has been tested by developers in Singapore.

The software platform with which scooters and cars are equipped is a self-learning system. Constant interaction with the environment improves the accuracy of the algorithms.



Another advantage of working with a unified platform is the ability to track free vehicles without the need to integrate a unified monitoring program into software platforms of various types. Free cars or scooters themselves declare their availability, and this information is displayed on the user device.


The MIT developers were able to create a scooter with automatic control in just two months. Of course, as mentioned above, the software for cars has already been created and worked, it remained only to adapt to a new type of vehicle.

Autonomous scooters were tested by volunteers. About a hundred people expressed their desire to test the system. Immediately before and after the tests, these people were asked to rate the safety of the devices on a scale from 1 to 5, where 5 is a feeling in the maximum system security. On average, the score grew from 3.6 points (before the tests) to 4.6 (immediately after them). According to the developers, this indicates that, in general, the development was successful.

The results of their work were presented by the project participants at the IEEE International Conference on Intelligent Transportation Systems ( IEEE International Conference on Intelligent Transportation Systems ), which was held last week in Rio de Janeiro, Brazil.