| Summary: | Due to the rising demand for rapid customisation in the manufacturing industry, Industry 4.0 and collaborative robots (or cobots) have increasingly been more relevant to the industry. For modern robots to be effective in today’s world, better human-robot communication must be established and implemented in a factory setting. In order to improve human-robot communications, two technologies can be leveraged to provide a solution i.e. Augmented Reality (AR) and Speech Recognition. This project concerns the testing and implementation of a speech recognition system to be used in an Augmented Reality cobot application. In particular, software packages are used to program a digital application that recognizes keywords and phrases. These keywords and phrases are either commands to be executed by a robot, or instructions for the robot to output certain data. Testing and experiments are conducted by repeating the keywords or phrases in varying conditions. The accuracy of the speech recognition under different situations are compiled. The data collected from the experiments are used as information to determine the most optimal settings for the proposed speech recognition solution. Findings show that noise affects speech recognition accuracy by up to 26%, while different words and phrases can affect speech recognition accuracy by up to 80%. Meanwhile, other factors have minimal impact on speech recognition accuracy. The speech recognition system is integrated with an AR prototype that the team members developed as a proof-of-concept on the use of AR and speech recognition for better cobot communication in a manufacturing setting.
|
|---|