High-performance image-guided continuum robots

传统的具有有限数量的串联刚性连杆的机器人已经不能成功地处理非结构化和受限环境中的复杂任务。例子包括许多微创手术应用，在受限环境中的搜索和救援操作，以及核工业和航空航天工业中的各种维护操作。或者，有人提出了连续体机器人(CRS)，由于它们的“连续主干结构”，为环境提供了一个“顺应”的界面。因此，CRS与传统机械手相比具有许多优点，如成本较低、在受限环境中的安全可操作性以及对环境接触条件的变化的适应性。因此，CRS有望对机器人在各个领域的成本和适用性产生重大影响。然而，对自主式CRS的研究还比较新，仍有许多问题需要解决，以克服它们的缺点，如刚度低、载荷小、精度低、可操作性差、工作空间和灵活性往往有限。*我们最近提出了合作CRS(CCRS)的新概念，不仅克服了CRS的上述局限性，而且吸收了可重构性，以增强其对新任务的适应能力。然而，自主CCR的许多理论和实践方面在很大程度上仍未得到处理。以基于图像的CCRS技术为重点，我们希望通过合并我们正在进行的关于(I)基于视觉的机器人控制和(Ii)CR建模和控制的协同研究，在CCRS的基础上取得重大进展。该计划的目标包括开发和验证：(1)先进的运动静态模型；(2)基于图像的实时形状和力传感的新方法；(3)稳健和通用的视觉伺服控制技术；(4)第一个稳健和主动的非刚性运动结构框架；以及(5)控制结果在医疗和安全防御应用的机器人辅助干预中的使用。*拟议的多学科研究计划预计将产生首个用于CCRS的运动学建模、传感和控制技术，实现许多重要功能，如提高导航的效率和准确性，改善灵活性和刚性，以及增强在非结构化和动态环境中操作的健壮性。这些技术将开辟大量新的机器人应用设计机会，支持加拿大航空航天、国防、服务、制造、能源和医疗系统等行业。此外，这项研究将带来新的研究方向，有利于机器人学、控制、计算机视觉、生物医学工程和自主系统等交叉领域的研究。

Conventional robotic manipulators with finite-number of serially-connected rigid links have not been successful in tackling complex tasks in unstructured and confined environments. Examples include many minimally-invasive surgery applications, search and rescue operations in confined environments, and various maintenance operations in nuclear and aerospace industries. Alternatively, continuum robots (CRs) have been proposed that, due to their “continuous backbone structures,” present a “compliant” interface to the environment. Thus CRs offer a number of advantages over traditional manipulators, such as lower cost, safe maneuverability in confined environments, and adaptability to variations in environmental contact conditions. Therefore, CRs are expected to have a significant impact on the cost and applicability of robots in various domains. However, research on autonomous CRs is fairly new and many issues remain to be addressed to overcome their shortcomings, such as low stiffness, small payload, low accuracy, poor manipulability, and often limited workspace and dexterity. ******We have recently proposed the new concept of cooperative CRs (CCRs) to not only overcome the aforementioned limitations of CRs but also assimilate reconfigurability for enhancing their adaptation capability to new tasks. However, many theoretical and practical aspects of autonomous CCRs are remaining largely untreated. With a focus on image-based techniques for CCRs, we expect to make significant advancements in the foundations of CCRs by merging our on-going synergetic research on (i) vision-based control of robots, and (ii) CR modeling and control. The objectives of this program include the development and validation of: (1) advanced kineto-static models; (2) novel methods for real-time image-based shape and force sensing; (3) robust and generic visual-servo control techniques; (4) the first robust and active non-rigid structure-from-motion frameworks; and (5) the use of the control results in robot-assisted interventions for medical and security-defence applications. ******The proposed multidisciplinary research program is anticipated to produce the first kinematic modeling, sensing, and control techniques for CCRs, enabling many important capabilities, such as enhanced efficiency and accuracy of navigation, improved dexterity and stiffness, and increased robustness of operation in unstructured and dynamic environments. These techniques will open up a host of new robot applications design opportunities, supporting Canadian industries in aerospace, defence, service, manufacturing, energy, and healthcare systems. Furthermore, this research will lead to new research directions and benefit overlapping research areas in robotics, control, computer vision, biomedical engineering, and autonomous systems.