Autonomous Systems

长期以来，机器人一直在工厂环境中工作，与人类分离，并在高度控制的工作场所执行重复性任务。然而，一个重大的转变正在发生。机器人正在走出工厂车间，在与人类共享的、不太可预测的工作场所工作。 挑战在于，机器人不再能够以开环方式执行路径或轨迹。 它们必须感知周围的环境，并实时做出反应。 典型的例子是自动驾驶。 我们不是在考虑重新设计自动驾驶汽车的道路，而是要求汽车在现有的基础设施内运行。 车辆必须持续感知周围环境，安全地通过施工区域，避免意外危险，甚至应对不稳定的驾驶员。这种转变需要新一代机器人能够自主智能地行动，并能够在现实环境中安全地执行任务。 Stephen L.博士Smith是加拿大自主系统研究主席，他正在致力于开发智能和规划算法，为下一代机器人提供动力。该主席的研究正在解决创造能够在以人为中心的环境中无缝操作的机器人的一些基本挑战。 史密斯的研究旨在创造具有几种关键能力的机器人。 机器人应根据用户的偏好调整其行为，并应在存在不确定性的情况下安全有效地运行。 这些挑战将通过开发算法来解决，这些算法使机器人能够：1）通过与用户的简单交互来学习如何执行任务; 2）在现实环境中规划安全有效的运动。以及3）主动识别不安全的情况，并将自主权无缝地移交给人类主管。这项研究的目标是创造可以直接集成到当前基础设施中的机器人，而不需要定制设施或对工作场所进行其他修改。这将极大地扩大自动化系统可用于提高效率和劳动生产率的行业范围，使农业，交付应用中的机器人车队，制造环境中的材料运输等应用成为可能。

Robots have long worked in factory environments, separated from humans and left to perform repetitive tasks in highly controlled workplaces. However, a major shift is taking place. Robots are moving beyond the factory floor to work in less predictable workplaces, shared with humans. The challenge is that robots can no longer execute a path or trajectory in open-loop. They must sense their environmental surroundings and react in real-time. The prototypical example is autonomous driving. Rather than redesigning roads with autonomous cars in mind, we are asking the cars to operate within the current infrastructure. Vehicles must continuously sense their surroundings, safely navigate through construction zones, avoid unexpected hazards, and even deal with erratic drivers.This shift is demanding a new generation of robots that act autonomously and intelligently and are capable of safely performing tasks in real-world environments. Dr. Stephen L. Smith, Canada Research Chair in Autonomous Systems, is working to develop the intelligence and planning algorithms that will power this next generation of robots.The chair's research is tackling some of the fundamental challenges in creating robots that can seamlessly operate in human-centric environments. Smith's research seeks to create robots that possess several key capabilities. Robots should adapt their behaviours to the preferences of their users, and should operate safely and efficiently in the presence of uncertainty. These challenges will be addressed by developing algorithms that enable robots to 1) learn how to perform tasks through simple interactions with a user; 2) plan safe and efficient motion in real-world environments.; and 3) proactively identify unsafe situations and seamlessly hand off autonomy to human supervisors.The goal of this research is to create robots that can be integrated directly into the current infrastructure, without requiring custom facilities or other modifications to the workplace. This would greatly expand the range of industries where autonomous systems can be used to improve efficiency and labour productivity, enabling applications from agriculture, robot fleets in delivery applications, material transport in manufacturing environments and beyond.