The Increasingly Fascinating Opportunity for Human-Robot-AI Interaction
The Increasingly Fascinating Opportunity for Human-Robot-AI Interaction
复制标题
人机人工智能交互的机遇日益诱人
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
M. Veloso
中科院分区：
文献类型：
--
作者：
M. Veloso
人机交互（HRI）和自主机器人技术仍然截然不同，但需要结合起来才能使这两个领域取得实质性进展。看到这种需求，我和我的团队一直在追求将移动服务机器人真正带入我们日常人类环境的道路。这个方向是一次重大的信念飞跃，也与我们在自主机器人领域的工作背道而驰，特别是对于蓬勃发展的机器人足球运动而言。与自动足球队不同的是，在自动足球队中，机器人在玩游戏时不允许获得输入或以任何其他方式与人类互动，而服务机器人本质上为与人类互动提供了机会和需求。结果是多个 CoBot 机器人在建筑物的室内移动。通过 CoBot 项目，我们学到了许多有关推动 HRI 和自主机器人技术研究的乐趣和陷阱的经验教训。人类在与高性能机器人的互动中扮演着多方面的角色，这给我们在 HRI 和自主机器人技术的交叉领域带来了许多意想不到的、令人着迷的问题，我们希望能够分享这些问题。 CoBot 机器人的基本目标有两个：接收用户的任务请求并在我们的建筑物中自主规划和执行此类任务。任务具有拾取和交付性质，要求机器人前往某个位置迎接访客或拾取物体，然后前往某个目的地位置接收访客或物体。最初，我们并行关注三个核心方向：（i）硬件、传感器、感知处理、电机控制、路径规划、安全导航和精确定位的平台问题； (ii) 协调多个协作机器人的用户请求的表示和调度； (iii) 与人类用户在特定任务上的互动，例如访客同伴。真正的机器人需要真正的工作才能将必需品到位。我们的工作不知疲倦地进行，直到我们能够完成从任务请求到自主的基本循环。
Human-robot interaction (HRI) and autonomous robotics remain worlds apart but need to come together for both fields to substantively advance. Seeing this need, my group and I have pursued the path of truly bringing mobile service robots to our daily human environments. This direction was a major leap of faith and a departure from our work in autonomous robotics, especially for the prospering robot soccer movement. As opposed to the autonomous soccer teams – in which robots are not allowed to get input or interact in any other way with humans while playing a game – service robots, by nature, open the opportunity and the need for interaction with humans. The result is several CoBot robots moving indoors in buildings. Through the CoBot project, we have learned many lessons about the joys and pitfalls of research that advances both HRI and autonomous robotics. The multifaceted role of humans in their interaction with highly capable robots has brought us to many unexpected and fascinating problems at the intersection of HRI and autonomous robotics, which we hope to share. The basic goal of the CoBot robots is twofold: to receive task requests from users and to autonomously plan and execute such tasks in our buildings. Tasks are of a pick-up and delivery nature, which require the robot to go to some location to greet a visitor or to pick up an object, and then go to some destination location for the visitor or for the object. Initially, we focused on the three core directions in parallel: (i) platform issues for hardware, sensors, sensory processing, motor control, route planning, safe navigation, and accurate localization; (ii) coordination for representing and scheduling of user requests for multiple CoBots; and (iii) interaction with human users on a specific task, such as a visitor companion. Real robots require real work to get the essentials in place. Our work proceeded tirelessly until we were able to complete the basic loop of going all the way from a task request to the autonomous