Robust Algorithm for Aquiring and Interpreting Contact Data via Active Robot Expansion

9361481 Townsend新一代机器人具有高带宽，由高性能无刷直流电机驱动的反向驱动传输，提供了改善力数据流回控制器的机会，而无需额外的力传感器。本研究将开发一种确定接触几何形状的算法，作为对控制器或遥控操作员的反馈。该算法将补充现有的力传感器和触觉传感器技术，后者提供了更高的精度，但在其他方面限制了系统的响应性和稳定性。该研究将编制一套力控制任务的原型，确定计算接触力矢量所需的特殊运动学方程，开发接触几何形状的数学和图形表示，确定所需速度的最佳实现算法，并概述第二阶段实验开发的规格。这项研究的结果将是更快、更灵巧的机器人，比现有的机器人系统传递更多的过程信息。美国公司，如Adept和Barrett Technology，已经帮助引领了国际社会在后驱机器人硬件的开发。通过改进控制算法来利用这些硬件改进，美国应该保持在机器人技术领域的竞争领先地位。

9361481 Townsend A new generation of robots with high-bandwidth, backdrivable transmissions driven by high-performance brushless DC motors presents the opportunity for improving the flow of force data back to the controller without additional force sensors. This research will develop an algorithm for determining contact geometry, to be used as feedback to the controller or teleoperator. The algorithm will complement existing force-sensor and tactile-sensor techniques which offer improved accuracy, but otherwise limit the responsiveness and stability of the system. The proposed research will compile a set of prototype force-control tasks, determine the special kinematic equations necessary to calculate contact force vectors, develop a mathematical and graphical representation of the contact geometries, determine the best implementation algorithm for the required speed, and outline specifications for experimental development in Phase II. The result of this research will be faster, more dexterous robots which relay greater information about the process than existing robot systems. US companies, such as Adept and Barrett Technology, have helped lead the international community in backdrivable robotic hardware development. By improving the control algorithms to capitalize on these hardware improvements, the US should retain its competitive lead in this area of robotics.