机器人学概述 / Introduction to Robotics

机器人之心：感知、规划与控制的科学

The Heart of Robots: The Science of Perception, Planning, and Control

---

机器人这个词，对于今天的我们而言，既熟悉又陌生。我们每天都能在新闻报道中听到它，在科幻电影中看到它，在工业生产线上观察到它，在智能家居设备中体验到它。然而，当被问及究竟什么是机器人、机器人学究竟研究什么时，许多人却难以给出一个准确而完整的答案。这种认知上的模糊性，恰恰说明了机器人学作为一个新兴学科的复杂性——它横跨机械工程、电子计算机科学、人工智能、控制理论、认知心理学、神经科学等多个领域，却又保持着自身独特的研究范式和方法论体系。

The term robot is both familiar and unfamiliar to us today. We hear about it in news reports, see it in science fiction films, observe it on industrial production lines, and experience it in smart home devices. However, when asked what exactly a robot is or what robotics research encompasses, many people struggle to provide an accurate and comprehensive answer. This cognitive ambiguity precisely reflects the complexity of robotics as an emerging discipline—it spans mechanical engineering, computer science, artificial intelligence, control theory, cognitive psychology, neuroscience, and many other fields, while maintaining its own unique research paradigm and methodological system.

本书正是为了回答这些根本性问题而撰写的。我们将从感知（Perception）、规划（Planning）和控制（Control）三个核心维度出发，系统性地阐述机器人学的基本原理、关键技术和未来发展方向。这一框架不仅是对机器人学知识的组织方式，更是对机器人本质的深刻揭示——真正的机器人之所以被称为“机器人”，正是因为它能够感知环境、规划行动并精确执行控制指令，从而在复杂多变的世界中自主完成各种任务。

This book is written precisely to answer these fundamental questions. We will systematically expound on the basic principles, key technologies, and future development directions of robotics from three core dimensions: Perception, Planning, and Control. This framework is not only a way of organizing knowledge about robotics but also a profound revelation of the essence of robots—true robots are called "robots" precisely because they can perceive the environment, plan actions, and precisely execute control commands, thereby autonomously completing various tasks in a complex and changing world.

---

1. 机器人的定义：从词源到本质

1. Definition of Robots: From Etymology to Essence

1.1 “Robot”一词的诞生

1.1 The Birth of the Word "Robot"

要理解什么是机器人，我们不妨从这个词的起源说起。“Robot”一词首次出现在1920年捷克作家卡雷尔·恰佩克（Karel Čapek）的戏剧《罗素姆万能机器人》（Rossum's Universal Robots）中。这个词的创造并非偶然，而是作者精心构思的结果——它源自捷克语中的“robota”，意为“强制劳动”或“苦役”。在恰佩克的设想中，机器人是一种被制造出来代替人类从事各种劳动的人造生物，它们外表与人类相似，却不具备情感和灵魂，存在的唯一目的就是服从命令、服务人类。

To understand what a robot is, we might as well start from the origin of this word. The term "robot" first appeared in 1920 in the play *Rossum's Universal Robots* by Czech writer Karel Čapek. The creation of this word was not accidental but a deliberate design by the author—it derives from "robota" in Czech, meaning "forced labor" or "drudgery." In Čapek's vision, robots were artificial creatures manufactured to replace humans in various labors. They resembled humans in appearance but lacked emotions and souls, existing solely to obey commands and serve humanity.

恰佩克的戏剧一经推出便引起轰动，“robot”一词也随之传播到全世界。然而，这部作品更深层的意义在于，它首次提出了一个人类至今仍在思考的问题：当机器能够完美替代人类劳动时，人类将何去何从？这个问题触及了机器人学的哲学根基——我们创造机器人，究竟是为了延伸自身的能力，还是为了制造自己的替代品？尽管一个世纪过去了，这个问题依然没有标准答案，但它始终指引着机器人学的研究方向，影响着每一项技术决策和伦理考量。

Once premiered, Čapek's play caused a sensation, and the word "robot" spread worldwide. However, the deeper significance of this work lies in its first raising a question that humanity is still contemplating today: when machines can perfectly replace human labor, where will humanity go? This question touches upon the philosophical foundation of robotics—do we create robots to extend our own capabilities, or to manufacture our own replacements? Although a century has passed and this question still has no standard answer, it has always guided the research direction of robotics and influenced every technical decision and ethical consideration.

1.2 机器人学的学术定义

1.2 Academic Definition of Robotics

在学术语境中，机器人学的定义经历了漫长的演变过程。1954年，英国发明家乔治·德沃尔（George Devol）申请了世界上第一台可编程工业机器人Unimate的专利，他在专利中将机器人描述为“一种用于执行通常由人类完成的操作的、能够编程的物料运输与操作设备”。这个定义强调了机器人的两个关键特征：可编程性和物料操作能力，至今仍是工业机器人领域最核心的技术指标。

In academic contexts, the definition of robotics has undergone a long evolution. In 1954, British inventor George Devol patented the world's first programmable industrial robot, Unimate, describing a robot as "a programmable material handling and manipulation device for performing operations normally done by humans." This definition emphasizes two key characteristics of robots: programmability and material manipulation capability, which remain the core technical indicators in the industrial robotics field.

1961年，Unimate正式在通用汽车公司的生产线上投入使用，标志着工业机器人从概念走向实际应用。这一历史事件不仅开启了自动化生产的新纪元，也为机器人学的学科体系奠定了实践基础。同期，日本机器人学家森政弘（Morihiro Mori）提出了著名的“恐怖谷理论”（Uncanny Valley），从心理学角度分析了人类对机器人外观的接受程度，为后来的仿人机器人设计提供了重要参考。

In 1961, Unimate was officially put into use on General Motors' production line, marking the transition of industrial robots from concept to practical application. This historical event not only opened a new era of automated production but also laid a practical foundation for the disciplinary system of robotics. During the same period, Japanese roboticist Morihiro Mori proposed the famous "Uncanny Valley Theory," analyzing human acceptance of robot appearances from a psychological perspective, providing an important reference for later humanoid robot design.

到了1980年代，随着计算机技术和人工智能的快速发展，机器人学的定义进一步扩展。IEEE（电气与电子工程师协会）在其技术标准中将机器人定义为“一种能够通过编程在两个或多个轴线上进行运动的、自动控制的、可重编程的、多用途的机械手”。这一定义在强调机械特性的同时，也开始关注机器人的智能属性和多功能特征。

By the 1980s, with the rapid development of computer technology and artificial intelligence, the definition of robotics expanded further. IEEE (Institute of Electrical and Electronics Engineers) defined a robot in its technical standards as "a reprogrammable, multifunctional manipulator designed to move materials, parts, tools, or specialized devices through variable programmed motions for the performance of a variety of tasks." This definition, while emphasizing mechanical characteristics, also began to focus on the intelligent attributes and multifunctional features of robots.

进入21世纪，机器人学的定义已经远远超越了传统的机械手臂和自动化设备。现代机器人学将机器人视为智能体（Agent）的一种形式，强调其自主性（Autonomy）、感知性（Perceptibility）和适应性（Adaptability）。美国机器人协会（RIA）将现代机器人定义为“一种可编程的、多功能的机械手，通过可变的运动来完成各种任务，以协助或替代人类工作”。这一定义虽然表述简洁，却涵盖了机器人最核心的本质特征：可编程性、多功能性、自动化运动和作业替代能力。

Entering the 21st century, the definition of robotics has far exceeded traditional mechanical arms and automated equipment. Modern robotics regards robots as a form of intelligent agent, emphasizing their autonomy, perceptibility, and adaptability. The Robotic Industries Association (RIA) defines a modern robot as "a programmable, multifunctional manipulator designed to move materials, parts, tools, or specialized devices through variable programmed motions for the performance of a variety of tasks, either to assist or substitute for human work." Although concise, this definition encompasses the most essential characteristics of robots: programmability, multifunctionality, automated motion, and capability for task substitution.

1.3 机器人的核心特征

1.3 Core Characteristics of Robots

综合以上各时期的定义，我们可以将机器人的核心特征归纳为以下几个方面：

Synthesizing definitions from various periods, we can summarize the core characteristics of robots as follows:

第一，物理实体性。 机器人必须具有物理形态，能够在物理世界中存在和运动。这使其与纯软件程序或虚拟智能体区分开来。尽管现代人工智能可以在数字世界中完成许多复杂任务，但严格意义上的机器人必须具备机械结构、传感器和执行器，能够与物理环境进行直接交互。这一特征将机器人限定在一个明确的范畴内：它既是信息的处理者，也是物质的操纵者。

First, physical embodiment. Robots must have a physical form and be able to exist and move in the physical world. This distinguishes them from pure software programs or virtual intelligent agents. Although modern artificial intelligence can accomplish many complex tasks in the digital world, robots in the strict sense must have mechanical structures, sensors, and actuators capable of directly interacting with the physical environment. This characteristic confines robots to a clear category: they are both information processors and material manipulators.

第二，感知能力。 机器人必须具备感知自身状态和外部环境的能力。这种能力通过各类传感器实现，包括视觉传感器（摄像头）、听觉传感器（麦克风）、触觉传感器（力/力矩传感器）、位置传感器（编码器）、距离传感器（激光雷达、超声波传感器）等。没有感知，机器人就无法获取执行任务