IDRT’s Gesture Recognition Research
design element to separate sections

Gesture Recognition

Through the use of the AcceleGlove™ and 3D depth-sensing camera technology, IDRT is currently developing gesture recognition ⁄ capture software and hardware for finger, hand, and arm movements.


How the Technology Works

The AcceleGlove™ is a nylon glove that uses six integrated sensors to measure motion and orientation of the hand, wrist, and fingers (18 degrees of freedom in total). The six sensors are all 3-axis MEMS accelerometers. One accelerometer is located on the back of each finger, and one is on the back of the hand. The signals are analyzed and recorded by the microcontroller. The recorded hand trajectory and finger positions are processed by the algorithms to determine the correct gesture, i.e. letter, word, or command. The microcontroller coverts the findings to ASCII characters. Last, the ASCII characters (output-text, speech, video, or command) can be sent to PC’s, tablets, smart phones, or smart televisions.



The primary application is as an input device for translating American Sign Language hand gestures to English text and speech. As part of a military or medical first responder communication system, for example, individuals can communicate with each other through standard hand signals without being in the line of sight of each other. The AcceleGlove™ can also capture hand or finger movements to act as commands for controlling consumer electronics, robots, and video games. The technology can be expanded to capture all body movements for sports training and physical therapy applications.


Next Stage, Add a Locator

The AcceleGlove™ cannot determine the position of a hand gesture relative to the body; hence the glove requires a “locator.” For example, the sign for the alphabet letter “n” is a fist with the thumb tucked under the index and middle fingers and signed to the side of the body or in front of the torso. However, if it is signed close to an individual’s jawline, the gesture means “niece.” If it is signed close to an individual’s temple, it means “nephew.” The addition of a camera to the glove enables the system to determine the position of the hand relative to the body and identify gestures more accurately.



How can data captured by the AcceleGlove™ be accessed?

  • In raw form, by reading a COM port. Can be accessed in Java by including a JAR file or in C++ by including a DLL file.

What is the AcceleGlove™ Software Development Kit (SDK) ⁄ Visualizer?

  • Open-sourced software that can read the glove input and visualize the data as a graph.

  • It can read multiple inputs for the same gesture and applies machine learning algorithms to “learn” a gesture.

  • It can learn a set of gestures and save them as (SQLite) gestural language database. The database can be easily integrated into any host application to make the gestural language accessible to it.

  • It is Java-based and requires Java Runtime Environment (JRE) to run.

What operating systems can I use with the AcceleGlove™?

  • Windows XP or Vista ⁄ Mac OS X (10.4 or higher, Intel-based) ⁄ Linux (Kernel 2.6.9 or higher).

  • Minimum of 512MB of RAM.

What are the orientation and acceleration ranges of the AcceleGlove™?

  • The orientation range is 180 degrees for each axis (x,y,z). The acceleration range is + ⁄ - 1.5g.

How sensitive are the AcceleGlove™ accelerometers?

  • The analog accelerometers measure data with a sensitivity of .0287 m ⁄ s2.

What is the resolution of AcceleGlove™ data?

  • The resolution is 10 bits A⁄D for each axis.

What is the sampling rate of the AcceleGlove™?

  • The maximum sampling rate is 35Hz (630 axes per second).


Intellectual Property

  • 7565295, Method and apparatus for translating hand gestures, issued July 21, 2009: Glove and arm link.

  • 20100063794, Method and apparatus for translating hand gestures, filed July 21, 2009: Glove and arm link.

  • 20100023314, ASL Glove with 3-Axis Accelerometers, filed August 8, 2007: 3-axis accelerometers

  • 61/620,182, Glove-Based Gesture Processing System, filed April 4, 2012: Glove ⁄ camera gesture recognition system.