Inspired by Alan Turing's idea of a child machine, 1 we introduce the formal definition of a robotic baby, an integrated 2 system with minimal world knowledge at birth, capable of 3 learning incrementally and interactively, and adapting to the 4 world. Within the definition, fundamental capabilities and system 5 characteristics of the robotic baby are identified and presented as 6 the system-level requirements. As a minimal viable prototype, the 7 Baby architecture is proposed with a systems engineering design 8 approach to satisfy the system-level requirements, which has beenverified and validated with simulations and experiments on a 10 robotic system. We demonstrate the capabilities of the robotic 11 baby in natural language acquisition and semantic parsing in 12 English and Chinese, as well as in natural language ground-13 ing, natural language reinforcement learning, natural language 14 programming and system introspection for explainability. The 15 education and evolution of the robotic baby are illustrated 16 with real world robotic demonstrations. Inspired by genetic 17 inheritance in human beings, knowledge inheritance in robotic 18 babies and its benefits regarding evolution are discussed. 19 Index Terms-Robotic baby, robotic architecture, instruction-20 based learning, natural language programming, semantic repre-21 sentation, explainable AI, evolutionary robotics. 22 I. INTRODUCTION 23 T HE idea starts with a baby. In 1950, Alan Turing had a 24 rough idea of a child machine, which, according to his de-25 scription, could be the starting point for an intelligent machine 26 to pass his test if under an appropriate course of education [1]. 27 Alan Turing's idea of a child machine could be developed as 28 follows. Let's suppose that the intelligent machine passing his 29 test has been created and it, by definition, could act just like 30 a normal human being. With such construction, the intelligent 31 machine has to start from somewhere, just like all human 32 beings starting from a baby. Indeed, the intelligent machine 33 has to start from being a baby as well. How should we bring 34 such a baby machine into life from Turing's rough idea? 35