Recently, intelligent robotics and automation have been widely used for various applications such as advanced driver assistance system (ADAS), autonomous driving technology, and robot control system. These require a real-time object detection as well as artificial intelligences (AI) such as path planning, cut in detection, behavior analysis, and localization. However, the heavy computation cost of those algorithms makes it difficult to realize real-time performance and low-power consumption especially for embedded systems. Such requirements motivate our research to design real-time ADAS processors integrating AI functions with low-power consumption. To achieve real-time intelligent ADAS functions, four different researches are carried out; Algorithm, Architecture, VLSI implementation, Humanistic AI.
   For algorithm research, we develop new ADAS software platform to obtain highly accurate depth map extraction and object detection with lower computation cost than conventional works. It exploits visual attention process of human brain so as to increase detection accuracy and speed by using depth and tracking information. Based on the new algorithm, a heterogeneous multi-core processor is designed with multiple-granularity parallelism for high throughput. After that, power and task management techniques, such as DVFS, power gating and dynamic resource management, are applied to the VLSI implementation of the processor. To realize humanistic AI on our SoC, we employ the way of human thinks, learns, understands, and predicts which overcome limitations of conventional A.I. including Deep Learning. It consists of the neuro-fuzzy algorithm, trajectory prediction, and inference. These studies are especially used for more advanced ADAS.