Aerial robotic contact-based inspection: planning and control
Aerial robotic contact-based inspection: planning and control
复制标题
DOI：
10.1007/s10514-015-9485-5
复制
发表时间：
2016-04-01
影响因子：
3.5
通讯作者：
Siegwart, Roland
中科院分区：
文献类型：
--
作者：
Alexis, Kostas;Darivianakis, Georgios;Siegwart, Roland
空中机器人接触式检测的挑战是本文的驱动动机。通过引入算法和控制定律，在控制和路径规划两个层面上解决该问题，确保通过接触和受控空中机器人物理交互进行最佳检查。针对飞行和物理交互稳定性，提出了一种混合模型预测控制框架，基于该框架，典型的四旋翼飞行器能够实现稳定主动的交互、环境表面上的精确轨迹跟踪以及力控制。凸优化技术使得这种控制器能够进行显式计算，该控制器考虑了自由飞行以及物理交互过程中的动态，确保了混合系统的全局稳定性，并在尊重车辆物理限制的同时提供最佳响应。该方案的进一步增强，允许在控制级别纳入最后手段的避障机制。依靠这样的控制律，开发了一种基于接触的检查规划器，它可以计算给定一组检查点内的最佳路线，同时避开环境表面上的任何障碍物或其他禁飞区。广泛的实验研究包括复杂的“空中书写”任务、与非平面和纹理表面的交互、执行多次检查操作和避障机动，表明了所提出方法的效率以及通过接触进行空中机器人检查的潜在能力。
The challenge of aerial robotic contact-based inspection is the driving motivation of this paper. The problem is approached on both levels of control and path-planning by introducing algorithms and control laws that ensure optimal inspection through contact and controlled aerial robotic physical interaction. Regarding the flight and physical interaction stabilization, a hybrid model predictive control framework is proposed, based on which a typical quadrotor becomes capable of stable and active interaction, accurate trajectory tracking on environmental surfaces as well as force control. Convex optimization techniques enabled the explicit computation of such a controller which accounts for the dynamics in free-flight as well as during physical interaction, ensures the global stability of the hybrid system and provides optimal responses while respecting the physical limitations of the vehicle. Further augmentation of this scheme, allowed the incorporation of a last-resort obstacle avoidance mechanism at the control level. Relying on such a control law, a contact-based inspection planner was developed which computes the optimal route within a given set of inspection points while avoiding any obstacles or other no-fly zones on the environmental surface. Extensive experimental studies that included complex "aerial-writing" tasks, interaction with non-planar and textured surfaces, execution of multiple inspection operations and obstacle avoidance maneuvers, indicate the efficiency of the proposed methods and the potential capabilities of aerial robotic inspection through contact.