Pursuing smart machines that recognize human's intend and event autonomously, we focus on the study of interactions between human and machines in mechanical, optical and electrical ways. This study includes researches on three-dimensional scene and event recognition for autonomous vehicle and robots (vision), voice recognition and high-dynamic range directional sound transducers (sound), and biometric sensors for monitoring of health condition by means of vibration and ultrasound transducers (touch). 3D scene recognition is a key technology toward commercialization of autonomous vehicle and smart robots. By means of this group's novel signal processing methodology combined with microsystems/ultrasound devices, high-definition, environmentally robust 3D scene recognition is enabled. This will provide accurate 3D recognition performance with compact, low-cost solution. The technical aspect of operating principle in view of mechanical signal processing will be highlighted and presented.