# BRI-003: Assistive Robot Teleoperation Using Behavior Trees > 本页由本地论文卡片生成。请不要直接编辑本页；修改源卡片后运行 `vp run docs:generate`。 > 中文支持说明：本页是中文站点镜像，保留论文标题、证据状态和来源字段的原始表述，避免未经核实的翻译改变证据边界。 > 内部 wiki 边界：本页提供论文卡片证据和用户提供的 PDF 下载入口，仅用于私有阅读；全文抽取文本不发布。 > 来源: `library/paper_cards/BRI-003.md` ## 论文访问 * 内部 PDF: 下载论文 PDF * DOI / 官方页面: [needs confirmation](https://arxiv.org/abs/2303.05177) * 开放访问页面: [Open access page](https://arxiv.org/abs/2303.05177) * 部署边界: 这些 PDF 链接只适合私有/受保护的内部 wiki。 ## BRI-003: Assistive Robot Teleoperation Using Behavior Trees ## Metadata * ID: BRI-003 * Title: Assistive Robot Teleoperation Using Behavior Trees * Year: 2023 * DOI / URL: https://arxiv.org/abs/2303.05177 * Local PDF: 见上方论文访问区块 * Text artifact: local-only path withheld from docs site * Review status: `extracted` ## Study Type * Track: Hybrid BCI / Brain-Robot Interface * Task: proposed shared-autonomy robot teleoperation framework using behavior trees for activities of daily living * Participants or dataset: no empirical participants or dataset are reported; the paper proposes future user and domain-expert studies * Hardware: generic robot arm teleoperation/controller setting; no named robot platform or BCI hardware is specified in the local text * Channels or sensors: generic low-dimensional user input and robot end-effector/world-state variables; BCI is mentioned as an example noisy input source, not as an implemented EEG protocol ## Methods * Paradigm: user defines an ADL as a PHAST behavior tree offline, then the tree guides teleoperation in real time by modifying user position commands toward predefined task trajectories * Signal processing or model: Phase Switching Teleoperation Behavior Trees with a fallback root, phase sequence nodes, condition nodes, and Shared Control Action Nodes that map user input plus current end-effector pose to a new pose * Training/calibration: no learning or BCI calibration is reported; reusable/predefined behavior-tree nodes and drag-and-drop-style interfaces are discussed as design enablers * Online/offline: offline activity description plus proposed real-time teleoperation execution; the paper provides a pouring case study but no completed online user experiment ## Results * Metrics: no quantitative task-success, workload, timing, or BCI decoding metrics are reported * Main findings: PHAST BTs are proposed as a modular, readable, extensible high-level control representation for shared-autonomy teleoperation, with a bottle-to-cup pouring example split into translation and rotation phases * Reported limitations: empirical verification is future work; hardware, sensing, BCI decoder, user population, and quantitative performance remain unspecified in the local text ## Relevance To This Project * Supports: shared-autonomy task execution can be represented as readable phase-based structures that guide low-dimensional or noisy user commands while preserving user control * Conflicts with: the paper does not implement SSVEP, MI, YOLO perception, grasp-pose estimation, or a validated BCI robotic grasping experiment * Design implication: behavior trees or phase templates may be useful for SAH-BRI-Grasp's robot execution layer after a target is selected, but this paper should not be cited as evidence for EEG decoding or scene-aware object selection ## Extracted Evidence | Claim | Status | Evidence Note | Page/Section | | --- | --- | --- | --- | | The paper motivates shared autonomy for noisy or delayed teleoperation inputs, including BCI as an example. | verified | The abstract states that noisy input signals such as BCI require assistive autonomy that keeps the user in control while following trajectories and avoiding obstacles. | Abstract | | PHAST BTs let a user describe an ADL offline and use the tree to assist real-time teleoperation. | verified | Figure 1 and the introduction describe offline activity definition and real-time command modification by the robot controller. | Introduction; Figure 1 | | Shared Control Action Nodes modify user input based on current robot/end-effector state. | verified | Section 3.1 defines SCANs as nodes that map user input and current pose to a new end-effector pose. | Phase Switching Teleoperation Behavior Trees; Shared Control Action Nodes | | The bottle-to-cup case study decomposes pouring into translation and rotation phases. | verified | The case study describes translation toward the cup, rotation around the bottle base, and rotation around the bottle neck. | Case Study | | The paper has not yet empirically verified the framework with user studies. | verified | The introduction and conclusion state that user/domain-expert experiments are planned future work. | Introduction; Conclusion and Outlook | | PHAST BTs are relevant to SAH-BRI-Grasp as a possible high-level robot task representation. | inferred | The paper supports shared-autonomy teleoperation structure, but not the project's specific SSVEP-MI-YOLO grasping loop. | Full paper | | Actual EEG hardware, SSVEP/MI protocol, detector, and robot platform details remain unverified. | needs confirmation | The local text discusses these only generically or not at all. | Full paper scope | ## Open Questions * Could SAH-BRI-Grasp express approach, pre-grasp, grasp, lift, and recovery as phase-based behavior-tree templates? * What robot platform, controller, and safety constraints would be needed to instantiate PHAST-like nodes in the project? * How should behavior-tree phase switching interact with BCI confirmations, cancellations, and no-control periods? * Does behavior-tree readability improve operator trust or workload in a measured BCI-grasping task?