Track-to-Track 센서 퓨전 방법을 이용한 고장 검출 알고리즘

Abstract

Recently, in the automotive industry, one of the most important mainstream is the intelligent vehicle. The intelligent vehicle acquires the information of the vehicle and its environments such as lane and a preceding vehicle. Then it provides convenience to the driver by using ADAS (Advanced Driver Assistance System) controller, for instance, LKA (Lane Keeping Assist), SCC (Smart Cruise Control), and FCA (Forward Collision-Avoidance Assist). It is important to get accurate sensing data of the vehicle and the environment. ISO26262 (“Road Vehicles- Functional safety”) is defined to address possible hazards of S/W and H/W for Automotive Electric/Electronic Systems. This paper deals with the FDI (Fault Detection and Isolation) algorithm to figure out the sensor faults of ADAS sensors and In-Vehicle Sensors such as IMU and Wheel Speed Sensors. Since it is hard to secure hardware redundancies of ADAS sensors, it creates underlying states of ADAS sensors and compares it with the outputs of the EKF composed of In-vehicle sensors. As the ADAS sensors, Front Camera, Left Camera, and Right Camera were selected.