MIT's Swarm Robotics Research

2019-11-01

A team from MIT has come up with a remarkable and simple solution to enable robots to achieve true cellular coordination: self-assembling robot cubes that can climb, rotate around each other, and jump in the air. These robots, which can interact with each other, have the potential to unlock hidden capabilities to accomplish complex tasks. After years of relentless effort, the team has made some progress. The robots can now use a barcode-like system on each face of the modules to "communicate," allowing them to recognize each other. A self-organized robot queue, consisting of 16 cubes, can now perform simple tasks or behaviors such as lining up, following arrows, and tracking lights. Each cube, referred to as an "M-block," contains an internal flywheel that moves at a speed of 20,000 revolutions per minute when the flywheel is braked. Additionally, each edge and face of the cube has a permanent magnet, allowing any two modules to connect to each other. Daniela Rus, a professor at MIT and director of CSAIL, explains that "M" stands for motion, magnet, and magic. Motion refers to the cubes' ability to move by jumping. Magnet refers to the cubes' ability to connect to other cubes using magnets, enabling them to move together and assemble structures. Magic refers to the fact that there are no visible power components on the cubes' exterior, making it appear as if they are driven by magic. The main author of a new paper on this system states, "What's unique about our approach is that it is cheap, robust, and scalable. What can be achieved with ten or a hundred modules may be limited, but what about ten thousand or a million?" In fact, as early as 2013, the team established the mechanism for M-Blocks and created six-faced cubes that could move using "inertial forces." When a cube module is placed on any face of another cube, each module can move in four basic directions, resulting in 24 different movement directions. Several algorithms have been proposed to assist robots in completing simple tasks. These algorithms led researchers to come up with the idea of a barcode-like system, where robot modules can perceive the identity of the other blocks they are connected to. This eliminates the need for modules to send signals to each other using infrared or wireless waves, which can cause interference and confusion. Finally, the team points out that building electronic products is highly challenging. To further develop the M-Block swarm, the team hopes that more robot research teams will conduct studies and manufacture larger swarms with the ability to form various structures.

RECENT AI NEWS

RECENT AI TOOLS