近年來，隨著自動駕駛越來越興盛，在汽車中可以找到各種感測器，像是聲納、影像、光達和雷達系統。這些感測器可以用來輔助駕駛，精確測量汽車前方、旁邊或後方物體的距離和相對速度，使駕駛在能見度差時或物體隱藏在盲點時，能得知物體的位置。如果感測器不能準確探測到目標，就會對駕駛員的安全構成嚴重威脅。相較於其他感測器，雷達具有不易受環境影響、量測距離長且精準的優勢。在汽車雷達中，調頻連續波（FMCW）雷達被廣泛使用，因為與脈衝雷達相比，其可降低訊號處理的硬體複雜度。然而實務上存在由干擾效應和反射損耗等引起的系統誤差，且這些誤差有可能被誤認為是所需訊號。如何有效的抑制雜訊以及如何準確判斷物體位置將是本篇論文探討的重點，首先考慮了線性回歸的方法將雜訊與訊號區隔並利用深層神經網路將物體定位。;With self-driving automobiles becoming more and more popular in recent years, various sensors could be found in cars such as Sonar, Vision, Lidar, and Radar Systems. These sensors are used to assist drivers. Exact measurement of distance and relative velocity of objects in front, besides, or behind the car allow the driver to perceive objects during bad visibility or objects hidden in the blind spot. If sensors do not accurately detect targets, it can pose a serious threat to driver’s safety. Compared with other sensors, radar has the advantages of not being easily affected by the environment, and being capable of measuring longer distances precisely. For automotive radar systems, Frequency Modulation Continuous Wave (FMCW) radar is generally utilized because the complexity of hardware in the signal processing part can be reduced, compared to that of pulse radar. However, there are interference effects, reflection loss and some system errors in the practical application, which could be mistaken for desired signals. How to reduce the noise and accurately position the object will be the focus of this paper. First, the linear regression method is used to distinguish the noise and the signal, and the deep neural network is used to locate the object.
