This paper presents the continuous work on a previous project on vision-based hand interaction design. The original work developed a vision-based approach of bare-hand-based posture recognition and motion tracking. This paper further investigates its application in virtual assembly. The accuracy and robustness of the developed approach enable product engineers to perform assembly operations while manipulating virtual objects directly in virtual environments with bare hands. The direct human involvement creates a user-defined assembly sequence, which contains the human knowledge of mechanical assembly. By extracting the precedence relationship of machinery parts, the system becomes capable of generating alternative assembly sequences for robot reprogramming. Operation of the presented approach is illustrated and analyzed with experiments.
展开▼