KinectBot: A Gesture Controlled Robot Developed In MIT

Front View [Source: SingularityHUB]
The ability of one MIT graduate student to hack the Kinect 3D sensor led to KinectBot, a device that uses iRobot products to generate 3D images of its surroundings and obey gesture commands.

According to a report posted on Singularity Hub, the device was developed by Philipp Robbel, a student at MIT’s Personal Robotics Group. He was able to hack into the Kinect 3D sensor with an iRobot platform and create a robot-like machine that will be able to see and understand its environment, as well as obey commands from humans.

Tentatively, it is still called the KinectBot, although Robbel can probably develop it further and name it after him. According to Robbel, his main goal in hacking the Kinect 3D sensor is to create a team of robots that can detect trapped people during search-and-rescue operations. The robots can supposedly work together during the operations.

Currently, Robbel has four of iRobot’s Create machines that he calls iPucks. He also has four quadrotors, although the Kinect sensors won’t be attached to that because of its weight.

A Proof of Concept

However, the KinectBot is not ready to be in mass production just yet. It’s a proof of concept, which means that Robbel is still tweaking and developing some aspects of the machine, so that he can get it to work on the function that he aimed it to be. According to the report, Robbel would need to perfect the 3D mapping feature first before we can see the rest of the results of the Kinect hacking.

Robbel is currently looking into developing the 3D mapping, so that the robots can help in detecting trapped people more efficiently and accurately.

A Tool for Disaster Relief

One can only imagine how this development in machines can help in the disaster relief alleviation. If Robbel succeeds, these machines could detect where people are and obey their commands. Although there are some risks of allowing the robots to obey what distressed people asked of them, Robbel said that finding one person could lead the robot to find the next victim using only the motions in which it found the first victim.

Open to Developers

Because KinectBot’s software and hardware are both hacked, the report said that it will also be easy for other tech experts to develop something like the machine. Robbel used the SLAM (Simultaneous Localization and Mapping) code, specifically Gmapping, and some visualizations from Mobile Robot Programming Toolkit (MRPT).

To make the KinectBot, he used some of his own interaction, human detection and gesture code. But all of these are open for download, so anyone can just download the software package and create machines through iRobot products.

Technology has wowed us several times, but can you imagine being saved by this robot (in case of accidents or natural disasters)? That is something that we have only seen in movies. But the idea is not that far-fetched already. With the proliferation of 3D images and gesture-controlled devices these days, the KinectBot is just one of the hundred prototypes that could forever alter the face of technology.

kinect bot1
Side View

Source: SingularityHUB