Low-Cost Wireless Intelligent Two Hand Gesture Recognition System

This paper elucidates the design and implementation of a low-cost wireless intelligent two hand gesture recognition system. The proposed system consists of a primary and secondary sub-system which work in tandem to recognize static gestures signed by the user. Each of the sub-systems consists of a sensory glove, embedded with custom made, low-cost flex and contact sensors interfaced with a ATMega328 microcontroller. The two sub-systems are wirelessly inter-connected through a pair of TI's CC2541 Bluetooth Low Energy (BLE) modules. The system recognizes gestures with the help of a Dual-mode Intelligent Agent (DIA) which operates in Identification Mode (IM) and intelligently switches to Enhanced Identification Mode (EIM) when the gesture fails to get recognized in IM. The EIM incorporates a Bit Stream Error Elimination (BSEE) algorithm which enhances the gesture recognition accuracy without the addition of any external hardware. The performance of the system was evaluated using a data set comprising of 196 static gestures from eight globally used sign languages. The system efficiency was found to be 80.06% in IM and was enhanced to 93.16% in EIM. The cost of the proposed system in prototype stage is USD 22, which the authors believe could be realized at under USD 10 on commercialization.