人的各項行動生活中幾乎離不開視覺,一旦發生視覺障礙,不僅平常生活起居,課業與工作困難重重,甚至寸步難行。因此,導盲輔具對於一般視障者為重要生活必需品。一般視障人士普遍使用白手杖(white cane),運用觸碰地面或是物體以確認前方是否可以行走以及障礙物的位置,點狀偵測障礙物有漏洞,而白手杖不可及的地方則一概無從得知。因此希望開發一系統,能夠擷取環境資訊,即時障礙物偵測之行動輔具,藉以改善視障者行動上之不便。 本研究將自走車架構與白手杖做一連接,藉由系統避障機制自動導引視障者至安全路徑。系統的特色為利用MSP430F149為主控體,TMS320VC5510 DSP單晶片為運算核心,並在自走車架構中搭配影像感測器與超音波感測器進行地面障礙物與懸空之障礙物偵測,於行進間進行即時避障,避障完成後回到原路徑繼續前進。 所完成的系統以不同大小、數量、顏色、懸空狀態的障礙物放置於三種不同的環境中進行避障測試。經實測,本系統大約距障礙物1公尺內即可即時偵測以及閃避障礙物其成功率高達92.1%以上,對於60-80公分高的懸空障礙物、照度為15~794 Lux的障礙物的偵測以及閃避亦有90%以上的成功率。雖然,系統並非如預期於1.5公尺處即偵測障礙物並閃避,究其原因在於各種物質對於超音波反射與吸收性質不同所造成。然而此結果仍然驗證了本研究之設計概念。
Human can’t do anything without vision in our daily life. Once there is a problem in our vision, it is very difficult to move from point A to point B, so much as to accomplish any daily activities at home, at school or at work. Therefore, travel aids are important requirements for visual impaired people. White cane, which is commonly used by visual impaired, guides the right of way and detects obstacles position by knocking on the ground or obstacles. However, there is limitation to this ‘dot detection’. Moreover, white cane doesn’t detect anything outside its reach. For this reason, it is proposed to develop a system that can retrieve environmental information and detect barricades promptly to improve traveling convenience for visual impaired. The proposed system combines autonomous vehicle and white cane to guide visual impaired automatically to safe routes by path planning. The center of the proposed system is a MSP430F149 microcontroller. A TMS320VC5510 DSP chip is used as the image computational center in this study. We use autonomous vehicle that combines CMOS image sensor and ultrasound sensor to detect obstacles either on the ground or hung in the air. After completing dodging the obstacle, the system returns to the original path and continuous moves forward. The completed system was tested using obstacles of different size, amount, color and suspension in three different environments. The test results indicate that the proposed system can detect and perform dodging maneuver within one meter to the obstacle. The success rate is higher than 92.1%. On the other hand, for obstacles hanging 60-80 cm above the ground and illuminative range from 15-794Lux, the detection and dodging rate is higher than 90%. Although, the proposed system doesn’t perform as our expectations to detect and dodge at 1.5 meters due to the difference in ultrasound reflection and absorbability of different material. The system can detect obstacle and dodge within 1 meter which proves the feasibility of the proposed study.