EAGER: Behavioral Repertoires for Soft Robotics

软机器人是一种引人注目的新平台，可以在非结构化、崎岖和动态的环境中与人类一起操作。然而，到目前为止，很少有软机器人可以在现场部署在搜救和灾难应对等场景中。这在一定程度上是由于寻找让软机器人有效移动的方法的挑战。这个项目的中心目标是建立方法，通过这些方法，软机器人可以在对自己的能力或周围环境知之甚少的情况下自主开发特定于环境的任务指令集。这些新技术将使机器人能够在受损或任务环境发生变化时快速有效地重新训练自己。重要的是，这项工作还将建立一个模式，让本科生作为独立研究人员参与和发展软机器人领域的高风险、高回报领域，从而扩大研究人员社区，并降低下一代机器人研究人员的进入门槛。具体地说，该项目将使用质量多样性算法高效自主地发现有效的软机器人行为，使它们能够在复杂环境中稳健和自适应地移动。这些技术将使用低成本、动态复杂的张拉整体机器人来开发。这项研究的具体目标是深入了解软机器人如何自主探索其能力范围，产生充分利用其动态的多模式行为指令集，并开发这些机器人能够稳健而高效地调整其指令指令以应对破坏和意外环境变化的方法。在整个过程中，这项工作将涉及使用高速、高分辨率运动捕获系统进行大量基于硬件的验证和测试。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Soft robots are a compelling new platform for operating alongside humans in unstructured, rugged, and dynamic environments. However, as of yet, very few soft robots are field-deployable in scenarios such as search-and-rescue and disaster response. This is due in part to the challenge of finding ways of making soft robots move effectively. The central aim of this project is to establish methods by which soft robots can autonomously develop environment-specific task repertoires with little or no prior knowledge about their own abilities or the surrounding environment. These new techniques will allow robots to quickly and efficiently retrain themselves when they are damaged or when their task environment changes. Importantly, this work will also establish a model for involving and developing undergraduate students as independent researchers in the high risk, high payoff field of soft robotics, thereby growing the community of researchers and lowering the barriers of entry for the next generation of robotics researchers.Specifically, this project will use of Quality Diversity Algorithms to efficiently and autonomously discover effective soft robotic behaviors that allow them to robustly and adaptively move in complex environments. These techniques will be developed using low-cost dynamically complex tensegrity-based robots. The specific goals of this research are to produce insights into how soft robots can autonomously explore the range of their abilities, producing multimodal repertoires of behaviors that fully leverage their dynamics, and to develop methods by which these robots can robustly and efficiently adapt their repertoires in response to damage and unexpected environmental change. Throughout, this effort will involve substantial hardware-based validation and testing using a high speed, high resolution motion capture system.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.