Constraint-based agents

这项研究计划的目标是发展建设智能系统的原则和实践。我们正在开发一种正式的理论，用于设计和实现基于约束的代理，包括机器人。我们的基于约束的代理（CBA）框架是一套理论和实用的工具，用于为嵌入在活动环境中的代理设计控制器。我们将在几个新的方向上发展这个模型。首先，我们计划将学习技巧融入CBA。其次，我们将允许与真人老师互动。第三，我们将研究多机器人区域探索算法。第四，我们计划结合各种技术细节来扩展模型的功能。我们为使用轮椅的人开发了一种位置感知辅助工具，支持在室内和室外环境中进行路线规划和导航。我们计划将导航系统与用户的个人调度系统结合起来，同时解决空间和时间的约束。我们最近还开发了一种智能机器人电动轮椅原型，它可以同时自我定位，利用航位推算、视觉和其他传感器，绘制其环境地图，并主动避开静态和动态障碍物。我们还打算将这两个系统集成到一个平台上。椅子使用者的初始目标人群包括长期护理机构中轻度至中度认知障碍的使用者。我们在温哥华和多伦多的合作者计划在不久的将来对这些系统进行试验。我们最近还开发了一个基于约束的城市规划系统，结合可持续性约束，使用多点触控桌面。我们计划扩展这个系统，以允许多个用户协作解决约束和一个自动约束求解器。CBA方法的科学新颖性源于其形式基础、实用性和作为混合智能系统作为代理的单一理论的简单性。实际工程意义在于，我们现在可以使用这些工具构建更安全、更可靠的基于约束的工件。

The goal of this research program is to develop the principles and practices of building intelligent systems. We are developing a formal theory for the design and implementation of constraint-based agents, including robots. Our Constraint-Based Agent (CBA) framework is a set of theoretical and practical tools for designing controllers for agents embedded in an active environment. We shall develop the model in several new directions. First, we plan to incorporate learning techniques into CBA. Second, we will allow interaction with a human teacher. Third, we will work on algorithms for multi-robot area exploration. Fourth, we plan to incorporate various technical elaborations to extend the power of the model. We have developed a location-aware aid for wheelchair-using individuals that supports route planning and navigation in indoor and outdoor environments. We plan to couple the navigation system with the user's personal scheduling system, solving both spatial and temporal constraints together. We have also recently developed a prototype intelligent robotic power wheelchair that can simultaneously self-localize, using dead-reckoning, vision and other sensors, map its environment and actively avoid static and dynamic obstacles. We also intend to integrate these two systems on a single platform. The initial target population of chair users consists of mild-to-moderately cognitively impaired users in long-term care facilities. Trials for these systems are being planned for the near future, with our collaborators in Vancouver and Toronto. We have also recently developed a constraint-based system for urban planning, incorporating sustainability constraints, using a multi-touch table surface. We plan to extend this system to allow collaborative solution of the constraints by several users and an automated constraint solver. The scientific novelty of the CBA approach stems from its formal foundation, its practicality and its simplicity as a unitary theory of hybrid intelligent systems as agents. The practical engineering significance is that we can now build safer and more reliable constraint-based artefacts using these tools.