This review provides an overview of recent advances in physical reservoir computing (RC), a computational framework suitable for temporal/sequential data processing. RC is derived from recurrent neural network models such as echo state networks and liquid state machines. The key advantage of RC is its fast learning and low training cost, achieved by training only the readout weights with simple methods like linear regression. Physical RC, which uses physical systems, substrates, and devices as reservoirs, has gained increasing attention due to its potential for hardware implementation and low learning costs. The review classifies physical RC systems based on the type of physical phenomenon used as reservoirs, including electronic, photonic, spintronic, mechanical, and biological systems. It discusses the characteristics and design principles of these reservoirs, highlighting their potential for developing next-generation machine learning hardware devices. The review also addresses current issues and future research directions in physical RC, emphasizing the need for efficient design principles and practical applications.This review provides an overview of recent advances in physical reservoir computing (RC), a computational framework suitable for temporal/sequential data processing. RC is derived from recurrent neural network models such as echo state networks and liquid state machines. The key advantage of RC is its fast learning and low training cost, achieved by training only the readout weights with simple methods like linear regression. Physical RC, which uses physical systems, substrates, and devices as reservoirs, has gained increasing attention due to its potential for hardware implementation and low learning costs. The review classifies physical RC systems based on the type of physical phenomenon used as reservoirs, including electronic, photonic, spintronic, mechanical, and biological systems. It discusses the characteristics and design principles of these reservoirs, highlighting their potential for developing next-generation machine learning hardware devices. The review also addresses current issues and future research directions in physical RC, emphasizing the need for efficient design principles and practical applications.