An inertial navigation system for inner-city ADAS

In this paper an inertial navigation system (INS) is presented that is designed to specifically meet the requirements of next generation inner-city driver assistance. It integrates information from wheel speed sensors, the steering wheel, as well as a low-cost inertial measurement unit. The INS is based on a two-staged processing. An Interacting Multiple Model (IMM) filter that combines a kinematic and a dynamic bicycle model first estimates the vehicle's motion in 2D. A subsequent filter integrates the 2D estimates with additional sensor data to yield reliable 3D position information. The proposed INS has been evaluated in test drives of several kilometers where it exhibited highly accurate relative positioning. The high accuracy in conjunction with the fact that the used sensors are already built in many today's production vehicles renders the presented INS attractive for driver assistance systems.