以網路語音助理及車載網路訊號發展之車內設備控制系統

Development of Voice Control System for Vehicle Equipment based on Voice Assistant and Controller Area Networks

吳建達(Jian-Da Wu) ； 林憲裕(Hsien-Yu Lin)

《車輛工程學刊》 13期 (2019/06) Pp. 1-20

摘要

本研究提出一種以網路語音助理及車載網路訊號為基礎的車內設備聲控系統，該系統可分為兩個部分；第一個部分是要瞭解透過車載網路傳送的訊號內容，此一部分包括使用逆向工程技術研究車載網路上傳輸訊號的含義；第二部分是透過人類的聲音命令，將訊號經由車載網路發送到汽車的電子設備系統上，這包括使用谷歌網路語音助理軟體開發套件在樹莓派上設計和建立語音辨識系統，並使用MCP2515車載網路模塊連接到樹莓派來發送車載網路訊號，和一個使用Python開發的關鍵字觸發程式來聯結它們。實驗結果證明，該系統能夠在實驗車輛上透過駕駛員的聲音來控制雨刷、大燈、電動車窗、車內空調等車內相關設備此外，在此實驗工作中，並提出了一種可以降低錯誤拒絕率並提升語音識別成功率的方法，以改善系統之控制效果。

關鍵字

車載網路 ； 語音辨識 ； 樹莓派 ； 谷歌網路語音助理 ； 錯誤拒絕率

並列摘要

This paper proposes a voice control system for vehicle electrical equipment based on the controller area networks (CAN) bus. This system architecture is divided into two parts. The first part is what to send into the CAN bus. This includes reverse engineering the meaning of data be transmitted in the CAN bus. The second part is how to send CAN bus signals into the in-vehicle networks using the human voice. This includes speech recognition design and implementation on a Raspberry Pi (RPI) using the Google assistant software development Kit (SDK) as well as a MCP2515 CAN bus module connected to RPI to send CAN bus signals. A keyword response application was developed in the Python environment to connect each other. The experimental results indicated that the proposed system could achieve windshield wiper, headlight, power windows and air conditioning control using the driver's voice commands in an experimental vehicle. In the final experimental work, a method for reducing the false rejection rate (FRR) is studied to improve the speech recognition rate.

並列關鍵字

controller area networks ； speech recognition ； Raspberry Pi ； Google Assistant ； fault reject rate