Embedded electronic systems used in vehicles are becoming more exposed and thus vulnerable to different types of faults and cybersecurity attacks. Examples of these systems are advanced driver assistance systems (ADAS) used in vehicles with different levels of automation. Failures in these systems could have severe consequences, such as loss of lives and environmental damages. Therefore, these systems should be thoroughly evaluated during different stages of product development. An effective way of evaluating these systems is through the injection of faults and monitoring their impacts on these systems. In this paper, we present SUFI, a simulation-based fault injector that is capable of injecting faults into ADAS features simulated in SUMO (simulation of urban mobility). Simulation-based fault injection is usually used at early stages of product development, especially when the target hardware is not yet available. Using SUFI we target car-following and lane-changing features of ADAS modelled in SUMO. The results of the fault injection experiments show the effectiveness of SUFI in revealing the weaknesses of these models when targeted by faults and attacks.