Recently, collaborations between tech companies in the field of autonomous driving and traditional automobile manufacturers have been actively carried out. Through these collaborations, autonomous driving technology is no longer in the research stage, but it is entering the first stage of mass production. The integration of autonomous driving systems and vehicle systems has given new challenges to vehicle systems that have not changed much over a long time, one of those conservative systems is the power distribution system. In this paper, an electrified power distribution(PD) system was introduced to respond to the functional safety requirements of a level 4 autonomous driving system. The most important requirement for the PD system that supplies 12V electric power to the electrical loads constituting autonomous driving systems is that "unintentional disruption of power supply shall not occur". To this end, the power distribution system should be able to monitor the voltage state of the redundant power supply network, control the power supplied to each electric load, and diagnose the power supply state in real-time. These diagnostic results are important information for HPC (High-Performance Computer) to accurately determine the type of failure that occurs in the electrical grid and establish an autonomous driving strategy accordingly. In particular, the power distribution system should maintain the power supply as far as possible even if an internal fault is issued, so that important sensors such as radar and lidar can maintain their operations. To meet these requirements, the multi-channel PD controller named Smart PDU (Power Distribution Unit) was designed using an MCU(Micro Controller Unit) and electronic high-side switches(IPS, Intelligent Power Switch). The 32-bit MCU monitors the input voltage and performs its operations to protect the Smart PDU itself and the electric loads connected thereto under low-voltage and high-voltage conditions. In addition, the Smart PDU is capable to measure the current consumption of each channel. A warning signal was generated for abnormal current consumption of each load according to the measured value, and actions are performed to cope with critical fault situations such as wire short. All of these operational states and measured values are transmitted to HPCs connected to Smart PDU over the CANFD(Controller Area Network Flexible Datarate) network so the HPC is able to monitor the operational states in real-time, and each channel can be controlled according to various service modes and driving strategies of autonomous vehicles. In particular, redundant internal PMIC (Power Management IC) was configured to prepare for the single PMIC failure, and a latch circuit was designed to maintain the operating state of the high-side switch to maintain the power supply state even in severe failure situations such as H/W fault or watch-dog. Through quantitative analysis such as FMEDA(Failure Mode Effects and Diagnostic Analysis), it was confirmed that Smart PDU has an ASIL-B level of robustness. Using the Smart PDU designed by the method described above, it is possible to maintain power supply to important electrical loads, such as radar and lidar, even in the case of a failure on the electrical power-net or the PD system. HPC is became capable to establish appropriate autonomous driving strategies according to the type and severity of the failure and secure the safety of passengers and pedestrians by receiving the operating state of the Smart PDU, power consumption information of each load, and failure state. Despite the achievements accomplished through this study, there are still many topics to be studied in the PD system of autonomous vehicles. As autonomous vehicles with level 4 or higher are still being developed by adding autonomous driving systems to existing mass-produced vehicle models, so system-level optimization such as minimizing the number of junction blocks or optimizing the routing paths of wiring harnesses has not been considered enough. However, if the electrical/electronic architecture of vehicles is designed in consideration of integration with the autonomous driving system, these remained challenges are expected to be sufficiently overcome in the near future.
Mr. Soon Myung KWON, Senior Research Engineer, Hyundai Motor Company