Adaptive Feedforward Control Applied to Robotics

应用于机器人的自适应前馈控制

• 批准号: 8707838
• 负责人: Christopher Atkeson
• 金额: $ 7万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1987
• 资助国家: 美国
• 起止时间: 1987-09-01 至 1989-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8707838&HistoricalAwards=false
• 关键词: Adaptive; Feedforward; Control; Applied; Robotics

The goal of this research is to explore how robots might follow a trajectory more accurately with practice of that trajectory. The research focuses on an adaptive feedforward control, where a feedforward command is modified on the basis of previous performance errors. Proposed trajectory learning algorithms will be analyzed mathematically, evaluated through simulation, and implemented on the MIT Serial Link Direct Drive Robot Arm. The objective of this research is to make robots, and machines in general, move more accurately. This research will lead to improved robot performance, and the ideas and algorithms developed can be applied directly to the control of a wide variety of machines. Robots and other machines will be able to move more accurately at higher speeds, allowing faster throughput and greater productivity. In addition, more accurate performance will make it easier to program robots to do particular tasks, as programmers will no longer have to anticipate and compensate for dynamic errors.

这项研究的目的是探索机器人如何遵循一个 更准确的轨迹。 的 研究集中于自适应前馈控制，其中 前馈命令在先前性能的基础上进行修改 错误. 提出的轨迹学习算法将进行分析 数学上，通过模拟进行评估，并在 MIT串行链路直接驱动机器人手臂。本研究的目的是 是让机器人和一般的机器更精确地移动。 这 研究将导致机器人性能的改善， 开发的算法可以直接应用于广泛的控制， 各种机器。 机器人和其他机器将能够移动 在更高的速度下更精确，从而实现更快的吞吐量， 更高的生产力。 此外，更精确的性能将使 它更容易编程机器人做特定的任务，因为程序员将 不再需要预测和补偿动态误差。