snapshot: sync mint main from current workspace 2026-03-03 16:32:56 +0800

这个提交包含在:
Elvin
2026-03-03 16:32:56 +08:00
当前提交 908eb4bbbc
共有 677 个文件被更改,包括 167208 次插入 和 0 次删除

46
website/README.md 普通文件
查看文件

@ -0,0 +1,46 @@
# Evo-RL Project Website
This folder contains a static project webpage for Evo-RL.
## Files
- `index.html`: page structure and content
- `styles.css`: visual system and responsive layout
- `app.js`: chart rendering, reveal animation, interaction, language/theme switching
- `assets/images`: shared static images used by website and root README
- `assets/gifs`: shared GIF gallery used by website and root README
- `assets/videos`: optional local mp4 placeholders (currently not used by the visual gallery)
## Local Preview
From repo root:
```bash
cd website
python -m http.server 8000
```
Open `http://localhost:8000`.
## UI Controls
- Language switch: `EN` / `中` in the top-right corner
- Theme switch: `Day` / `Night` in the top-right corner
- Preferences persist via `localStorage` (`evorl_lang`, `evorl_theme`)
## Replace Visual Media
The website and root README now share media under `assets/`:
- `assets/images/*`
- `assets/gifs/*`
## Replace Curve Data
Edit `metrics` in `app.js`:
- `value.points`
- `advantage.points`
- `success.points`
Each array corresponds to rounds `R1 ... R8` in the chart.

622
website/app.js 普通文件
查看文件

@ -0,0 +1,622 @@
const chartCanvas = document.getElementById("signalChart");
const metricButtons = document.querySelectorAll(".metric-btn");
const langButtons = document.querySelectorAll(".lang-btn");
const themeToggle = document.getElementById("themeToggle");
const themeLabel = document.getElementById("themeLabel");
const themeIcon = document.getElementById("themeIcon");
const STORAGE_LANG_KEY = "evorl_lang";
const STORAGE_THEME_KEY = "evorl_theme";
const validLangs = ["en", "zh"];
const validThemes = ["dark", "light"];
const i18n = {
en: {
pageTitle: "Evo-RL | Continuous Open-Source Real-World RL",
pageDescription: "Evo-RL: continuous open-source real-world reinforcement learning on SO101 and AgileX PiPER, with a full reproducible CLI pipeline and offline value/policy training.",
navFocus: "Focus",
navPipeline: "Pipeline",
navResults: "Results",
navMedia: "Media",
navReproduce: "Reproduce",
trafficTitle: "Site Traffic",
trafficVisitors: "Visitors",
trafficViews: "Views",
heroEyebrow: "Continuous Open-Source Real-World RL",
heroSubline: "SO101 to AgileX PiPER",
heroSummary:
"A LeRobot-aligned, full-pipeline real-world offline RL workflow for real robots: collect rollout data, train value functions and policies offline, infer advantage tags, and iterate deployment in a closed loop.",
newsLabel: "NEWS",
newsDate: "February 26, 2026",
newsText: "First SO101 real-world RL baseline + reproducible CLI released.",
btnRepo: "GitHub Repo",
btnDocs: "LeRobot Docs",
statPlatforms: "Robot Platforms",
statOpenTasks: "Open Task Datasets",
statOpenModels: "Open Models",
statReleaseYear: "Release Year",
focusEyebrow: "Project Focus",
focusTitle: "Open Infrastructure for Real-World RL",
focus1Title: "Open real-world RL pipeline",
focus1Body:
"Runnable workflows on SO101 and AgileX PiPER covering collection, offline value/policy training, and iterative deployment.",
focus2Title: "Reproducible assets",
focus2Body:
"Codebase, model layout, dataset structure, and command-line templates are released to reduce setup friction and improve reproducibility.",
focus3Title: "Community co-evolution",
focus3Body:
"Reproduce existing real-world RL methods, extend value/policy variants, and continuously publish experiments and benchmarks for collective progress.",
methodEyebrow: "Method",
methodTitle: "6-Stage Closed-Loop Pipeline",
step1: "<strong>System preparation</strong> to ensure robot, sensors, and control stack are ready.",
step2: "<strong>Interactive data collection</strong> with human guidance and episode-level outcomes.",
step3: "<strong>Value modeling</strong> to estimate trajectory quality from accumulated data.",
step4: "<strong>Signal annotation</strong> to convert value outputs into usable training supervision.",
step5: "<strong>Policy update and deployment</strong> to run the improved policy in real tasks.",
step6: "<strong>Data refresh and iteration</strong> to continue improving robustness and success rate.",
resultsEyebrow: "Results",
resultsTitle: "Training Signals Across Iterations",
metricValue: "Value",
metricAdvantage: "Advantage",
metricSuccess: "Success Rate",
chartNote: "Illustrative trend view for project webpage. Replace with your run logs for paper-ready plots.",
insightTitle: "What This Visual Captures",
insight1: "Value trend stabilizes as rollouts and labels improve across rounds.",
insight2: "Advantage separates better trajectories used for ACP tagging.",
insight3: "Success rate rises with iterative deployment and data refresh.",
noteStrong:
"Current default ACP ratio in README pipeline: <code>--acp.positive_ratio=0.3</code>.",
mediaEyebrow: "Qualitative Demos",
mediaTitle: "Visual Results Gallery",
mediaTag1: "VALUE VISUAL",
mediaTitle1: "Success and Failure Case",
mediaBody1: "Frame-level value visualization on real robot rollouts.",
mediaTag2: "POLICY ROLLOUT",
mediaTitle2: "Deployment Visual Results",
mediaBody2: "Final policy behavior clips in target task execution.",
mediaTag3: "HUMAN-IN-THE-LOOP",
mediaTitle3: "Intervention and Recovery Results",
mediaBody3: "Human guidance segments and autonomous recovery outcomes.",
architectureEyebrow: "Architecture",
architectureTitle: "LeRobot-Aligned VLA + Value Loop",
architectureBody1:
"Evo-RL reuses LeRobot's practical data and inference stack while adding value-guided filtering signals (<code>value</code>, <code>advantage</code>, <code>acp_indicator</code>) to improve iterative policy training.",
architectureBody2:
"The same design supports single-GPU and accelerate-based multi-GPU workflows for both value and policy stages.",
relatedEyebrow: "Related Work",
relatedTitle: "Context in Open Robot Learning",
reproduceEyebrow: "Reproducibility",
reproduceTitle: "Core CLI Commands",
humanInloopRecord: "Human-in-loop Record",
valueTrain: "Value Training",
policyTrain: "Policy Training",
communityEyebrow: "Model & Community",
communityTitle: "Links and Contact",
releaseStatus: "Release Status",
release1: "Hugging Face model: coming soon",
release2: "Hugging Face dataset: coming soon",
release3: "Canonical version tags will be pinned once published",
wechatTitle: "WeChat Group",
wechatDesc: "Scan to join EvoMind's WeChat community and follow project updates.",
wechatContact: "Contact",
citationEyebrow: "Citation",
copyBib: "Copy BibTeX",
copied: "Copied",
copyFailed: "Copy Failed",
themeDark: "Night",
themeLight: "Day",
themeToggleAria: "Theme toggle",
chartLatest: "Latest"
},
zh: {
pageTitle: "Evo-RL | 持续开源真实世界强化学习",
pageDescription: "Evo-RL:面向 SO101 与 AgileX PiPER 的真实世界强化学习开源项目,提供可复现的全流程命令行管线,并包含离线 value/policy 训练。",
navFocus: "项目目标",
navPipeline: "方法流程",
navResults: "实验结果",
navMedia: "视频展示",
navReproduce: "复现命令",
trafficTitle: "访问统计",
trafficVisitors: "访客人数",
trafficViews: "查看次数",
heroEyebrow: "持续开源的真实世界强化学习",
heroSubline: "从 SO101 到 AgileX PiPER",
heroSummary:
"基于 LeRobot 实践体系的真实机器人全流程离线 RL:采集 rollout 数据、离线训练 value/policy、推理 advantage 标签,并在闭环中持续迭代部署。",
newsLabel: "最新进展",
newsDate: "2026年2月26日",
newsText: "首个 SO101 真实世界 RL baseline 与可复现 CLI 工作流已发布。",
btnRepo: "GitHub 仓库",
btnDocs: "LeRobot 文档",
statPlatforms: "机器人平台",
statOpenTasks: "社区开源任务数据",
statOpenModels: "社区开源模型",
statReleaseYear: "发布日期",
focusEyebrow: "项目目标",
focusTitle: "真实世界 RL 的开源基础设施",
focus1Title: "开放真实世界 RL 全流程",
focus1Body: "在 SO101 与 AgileX PiPER 上提供可运行流程,覆盖采集、离线 value/policy 训练与迭代部署。",
focus2Title: "可复现资产",
focus2Body: "持续发布代码、模型组织方式、数据结构与命令模板,降低复现实验门槛。",
focus3Title: "社区共演化",
focus3Body: "复现已有真实世界 RL 方法,扩展 value/policy 变体,并持续发布实验与基准推动社区进步。",
methodEyebrow: "方法",
methodTitle: "6阶段闭环流程",
step1: "<strong>系统准备</strong>:确保机器人、传感器与控制链路可稳定运行。",
step2: "<strong>交互式数据采集</strong>:通过人类引导采集轨迹并记录回合结果。",
step3: "<strong>价值建模</strong>:基于当前数据学习轨迹质量评估能力。",
step4: "<strong>信号标注</strong>:将价值输出转化为可用于训练的监督信号。",
step5: "<strong>策略更新与部署</strong>:训练改进后的策略并投入真实任务执行。",
step6: "<strong>数据刷新与迭代</strong>:持续循环优化鲁棒性与任务成功率。",
resultsEyebrow: "结果",
resultsTitle: "跨轮次训练信号",
metricValue: "Value",
metricAdvantage: "Advantage",
metricSuccess: "成功率",
chartNote: "当前为示意曲线。替换为你的真实训练日志后可直接用于论文或展示。",
insightTitle: "这张图反映了什么",
insight1: "随着轮次增加和标注改进,Value 曲线逐步稳定。",
insight2: "Advantage 更好地区分高质量轨迹,用于 ACP 标签化。",
insight3: "在闭环部署和数据刷新下,任务成功率持续提升。",
noteStrong: "README 默认 ACP 比例:<code>--acp.positive_ratio=0.3</code>。",
mediaEyebrow: "定性展示",
mediaTitle: "可视化结果画廊",
mediaTag1: "VALUE 可视化",
mediaTitle1: "成功与失败案例",
mediaBody1: "真实机器人 rollout 的逐帧 value 可视化。",
mediaTag2: "策略 Rollout",
mediaTitle2: "策略部署可视化结果",
mediaBody2: "最终策略在目标任务中的执行片段。",
mediaTag3: "人类在环",
mediaTitle3: "干预与恢复结果",
mediaBody3: "展示人工引导片段与自主恢复结果。",
architectureEyebrow: "架构",
architectureTitle: "LeRobot 对齐的 VLA + Value 闭环",
architectureBody1:
"Evo-RL 复用了 LeRobot 的实用数据与推理栈,并加入 value 引导过滤信号(<code>value</code>、<code>advantage</code>、<code>acp_indicator</code>)以提升迭代训练效果。",
architectureBody2: "同一套设计同时支持单卡与 accelerate 多卡的 value/policy 训练流程。",
relatedEyebrow: "相关工作",
relatedTitle: "开源机器人学习语境",
reproduceEyebrow: "复现",
reproduceTitle: "核心 CLI 命令",
humanInloopRecord: "Human-in-loop 采集",
valueTrain: "Value 训练",
policyTrain: "Policy 训练",
communityEyebrow: "模型与社区",
communityTitle: "链接与联系",
releaseStatus: "发布状态",
release1: "Hugging Face 模型:即将发布",
release2: "Hugging Face 数据集:即将发布",
release3: "发布后会在此固定 canonical 版本标签",
wechatTitle: "微信交流群",
wechatDesc: "扫码加入 EvoMind 微信社区,获取 Evo-RL 的最新项目动态。",
wechatContact: "联系邮箱",
citationEyebrow: "引用",
copyBib: "复制 BibTeX",
copied: "已复制",
copyFailed: "复制失败",
themeDark: "夜间",
themeLight: "白天",
themeToggleAria: "切换白天夜间模式",
chartLatest: "最新"
}
};
const metrics = {
value: {
labels: { en: "Mean Value", zh: "平均 Value" },
unit: "",
color: "#2af2d0",
fill: "rgba(42, 242, 208, 0.16)",
points: [0.41, 0.53, 0.69, 0.78, 0.92, 1.01, 1.09, 1.15]
},
advantage: {
labels: { en: "Mean Advantage", zh: "平均 Advantage" },
unit: "",
color: "#f8b84e",
fill: "rgba(248, 184, 78, 0.14)",
points: [0.07, 0.11, 0.14, 0.21, 0.24, 0.28, 0.3, 0.33]
},
success: {
labels: { en: "Task Success Rate", zh: "任务成功率" },
unit: "%",
color: "#7db6ff",
fill: "rgba(125, 182, 255, 0.15)",
points: [22, 29, 37, 46, 54, 61, 67, 73]
}
};
const rounds = ["R1", "R2", "R3", "R4", "R5", "R6", "R7", "R8"];
let activeMetric = "value";
let currentLang = validLangs.includes(localStorage.getItem(STORAGE_LANG_KEY))
? localStorage.getItem(STORAGE_LANG_KEY)
: "en";
let currentTheme = validThemes.includes(localStorage.getItem(STORAGE_THEME_KEY))
? localStorage.getItem(STORAGE_THEME_KEY)
: "light";
const textBindings = [
{ selector: "nav.site-nav a[href='#focus']", key: "navFocus" },
{ selector: "nav.site-nav a[href='#pipeline']", key: "navPipeline" },
{ selector: "nav.site-nav a[href='#results']", key: "navResults" },
{ selector: "nav.site-nav a[href='#media']", key: "navMedia" },
{ selector: "nav.site-nav a[href='#reproduce']", key: "navReproduce" },
{ selector: "#traffic .traffic-title", key: "trafficTitle" },
{ selector: "#traffic .traffic-item:nth-child(2) .traffic-label", key: "trafficVisitors" },
{ selector: "#traffic .traffic-item:nth-child(3) .traffic-label", key: "trafficViews" },
{ selector: ".hero .eyebrow", key: "heroEyebrow" },
{ selector: ".hero-left h1 span", key: "heroSubline" },
{ selector: ".hero-summary", key: "heroSummary" },
{ selector: ".news-pill span", key: "newsLabel" },
{ selector: ".news-pill strong", key: "newsDate" },
{ selector: ".news-pill em", key: "newsText" },
{ selector: ".hero-actions .btn-primary", key: "btnRepo" },
{ selector: ".hero-actions .btn-ghost", key: "btnDocs" },
{ selector: ".hero-stats .stat-card:nth-child(1) .stat-label", key: "statPlatforms" },
{ selector: ".hero-stats .stat-card:nth-child(2) .stat-label", key: "statOpenTasks" },
{ selector: ".hero-stats .stat-card:nth-child(3) .stat-label", key: "statOpenModels" },
{ selector: ".hero-stats .stat-card:nth-child(4) .stat-label", key: "statReleaseYear" },
{ selector: "#focus .section-heading .eyebrow", key: "focusEyebrow" },
{ selector: "#focus .section-heading h2", key: "focusTitle" },
{ selector: "#focus .card-grid .glass-card:nth-child(1) h3", key: "focus1Title" },
{ selector: "#focus .card-grid .glass-card:nth-child(1) p", key: "focus1Body" },
{ selector: "#focus .card-grid .glass-card:nth-child(2) h3", key: "focus2Title" },
{ selector: "#focus .card-grid .glass-card:nth-child(2) p", key: "focus2Body" },
{ selector: "#focus .card-grid .glass-card:nth-child(3) h3", key: "focus3Title" },
{ selector: "#focus .card-grid .glass-card:nth-child(3) p", key: "focus3Body" },
{ selector: "#pipeline .section-heading .eyebrow", key: "methodEyebrow" },
{ selector: "#pipeline .section-heading h2", key: "methodTitle" },
{ selector: "#pipeline .pipeline-list li:nth-child(1) p", key: "step1", html: true },
{ selector: "#pipeline .pipeline-list li:nth-child(2) p", key: "step2", html: true },
{ selector: "#pipeline .pipeline-list li:nth-child(3) p", key: "step3", html: true },
{ selector: "#pipeline .pipeline-list li:nth-child(4) p", key: "step4", html: true },
{ selector: "#pipeline .pipeline-list li:nth-child(5) p", key: "step5", html: true },
{ selector: "#pipeline .pipeline-list li:nth-child(6) p", key: "step6", html: true },
{ selector: "#results .section-heading .eyebrow", key: "resultsEyebrow" },
{ selector: "#results .section-heading h2", key: "resultsTitle" },
{ selector: ".chart-toolbar .metric-btn[data-metric='value']", key: "metricValue" },
{ selector: ".chart-toolbar .metric-btn[data-metric='advantage']", key: "metricAdvantage" },
{ selector: ".chart-toolbar .metric-btn[data-metric='success']", key: "metricSuccess" },
{ selector: ".chart-note", key: "chartNote" },
{ selector: ".insights-card h3", key: "insightTitle" },
{ selector: ".insights-card ul li:nth-child(1)", key: "insight1" },
{ selector: ".insights-card ul li:nth-child(2)", key: "insight2" },
{ selector: ".insights-card ul li:nth-child(3)", key: "insight3" },
{ selector: ".note-strong", key: "noteStrong", html: true },
{ selector: "#media .section-heading .eyebrow", key: "mediaEyebrow" },
{ selector: "#media .section-heading h2", key: "mediaTitle" },
{ selector: "#media .viz-lane:nth-child(1) .lane-kicker", key: "mediaTag1" },
{ selector: "#media .viz-lane:nth-child(1) h3", key: "mediaTitle1" },
{ selector: "#media .viz-lane:nth-child(1) .lane-note", key: "mediaBody1" },
{ selector: "#media .viz-lane:nth-child(2) .lane-kicker", key: "mediaTag2" },
{ selector: "#media .viz-lane:nth-child(2) h3", key: "mediaTitle2" },
{ selector: "#media .viz-lane:nth-child(2) .lane-note", key: "mediaBody2" },
{ selector: "#media .viz-lane:nth-child(3) .lane-kicker", key: "mediaTag3" },
{ selector: "#media .viz-lane:nth-child(3) h3", key: "mediaTitle3" },
{ selector: "#media .viz-lane:nth-child(3) .lane-note", key: "mediaBody3" },
{ selector: "#architecture .section-heading .eyebrow", key: "architectureEyebrow" },
{ selector: "#architecture .section-heading h2", key: "architectureTitle" },
{ selector: "#architecture .architecture-block div p:nth-child(1)", key: "architectureBody1", html: true },
{ selector: "#architecture .architecture-block div p:nth-child(2)", key: "architectureBody2" },
{ selector: "#related .section-heading .eyebrow", key: "relatedEyebrow" },
{ selector: "#related .section-heading h2", key: "relatedTitle" },
{ selector: "#reproduce .section-heading .eyebrow", key: "reproduceEyebrow" },
{ selector: "#reproduce .section-heading h2", key: "reproduceTitle" },
{ selector: "#reproduce .card-grid .glass-card:nth-child(1) h3", key: "humanInloopRecord" },
{ selector: "#reproduce .card-grid .glass-card:nth-child(2) h3", key: "valueTrain" },
{ selector: "#reproduce .card-grid .glass-card:nth-child(3) h3", key: "policyTrain" },
{ selector: "#community .section-heading .eyebrow", key: "communityEyebrow" },
{ selector: "#community .section-heading h2", key: "communityTitle" },
{ selector: "#community .card-grid .glass-card:nth-child(1) h3", key: "releaseStatus" },
{ selector: "#community .card-grid .glass-card:nth-child(1) ul li:nth-child(1)", key: "release1" },
{ selector: "#community .card-grid .glass-card:nth-child(1) ul li:nth-child(2)", key: "release2" },
{ selector: "#community .card-grid .glass-card:nth-child(1) ul li:nth-child(3)", key: "release3" },
{ selector: "#community .community-grid .qr-card h3", key: "wechatTitle" },
{ selector: "#community .community-grid .qr-card .qr-desc", key: "wechatDesc" },
{ selector: "#community .community-grid .qr-card .qr-contact-label", key: "wechatContact" },
{ selector: "#citation .section-heading .eyebrow", key: "citationEyebrow" }
];
function t(key) {
return i18n[currentLang][key] || i18n.en[key] || key;
}
function setTextContent(selector, value, useHtml = false) {
const el = document.querySelector(selector);
if (!el) return;
if (useHtml) {
el.innerHTML = value;
return;
}
el.textContent = value;
}
function applyTranslations() {
document.title = t("pageTitle");
const description = document.querySelector("meta[name='description']");
if (description) {
description.setAttribute("content", t("pageDescription"));
}
textBindings.forEach((binding) => {
setTextContent(binding.selector, t(binding.key), Boolean(binding.html));
});
if (themeToggle) {
themeToggle.setAttribute("aria-label", t("themeToggleAria"));
}
const copyButton = document.getElementById("copyBib");
if (copyButton) {
copyButton.textContent = t("copyBib");
}
}
function applyLanguage(lang) {
if (!validLangs.includes(lang)) return;
currentLang = lang;
document.documentElement.lang = lang === "zh" ? "zh-CN" : "en";
localStorage.setItem(STORAGE_LANG_KEY, lang);
langButtons.forEach((btn) => {
btn.classList.toggle("is-active", btn.dataset.lang === lang);
});
applyTranslations();
updateThemeToggle();
drawChart(activeMetric);
}
function updateThemeToggle() {
if (!themeLabel || !themeIcon) return;
const isDark = currentTheme === "dark";
themeIcon.textContent = isDark ? "◐" : "◑";
themeLabel.textContent = isDark ? t("themeDark") : t("themeLight");
}
function applyTheme(theme) {
if (!validThemes.includes(theme)) return;
currentTheme = theme;
document.body.dataset.theme = theme;
localStorage.setItem(STORAGE_THEME_KEY, theme);
updateThemeToggle();
drawChart(activeMetric);
}
function drawChart(metricKey) {
if (!chartCanvas) return;
const metric = metrics[metricKey];
const ctx = chartCanvas.getContext("2d");
const dpr = window.devicePixelRatio || 1;
const cssWidth = chartCanvas.clientWidth || 960;
const cssHeight = Math.max(320, Math.round(cssWidth * 0.5));
chartCanvas.width = Math.round(cssWidth * dpr);
chartCanvas.height = Math.round(cssHeight * dpr);
ctx.setTransform(dpr, 0, 0, dpr, 0, 0);
const style = getComputedStyle(document.body);
const chartBg = style.getPropertyValue("--chart-bg").trim() || "rgba(4, 10, 15, 0.95)";
const chartGrid = style.getPropertyValue("--chart-grid").trim() || "rgba(112, 153, 186, 0.25)";
const chartAxis = style.getPropertyValue("--chart-axis").trim() || "#93aec0";
const chartLatest = style.getPropertyValue("--chart-latest").trim() || "#d1e3ef";
const w = cssWidth;
const h = cssHeight;
const pad = { top: 28, right: 24, bottom: 42, left: 52 };
const plotW = w - pad.left - pad.right;
const plotH = h - pad.top - pad.bottom;
ctx.clearRect(0, 0, w, h);
ctx.fillStyle = chartBg;
ctx.fillRect(pad.left, pad.top, plotW, plotH);
const values = metric.points;
const min = Math.min(...values);
const max = Math.max(...values);
const range = max - min || 1;
const xStep = plotW / (values.length - 1);
const toX = (i) => pad.left + i * xStep;
const toY = (val) => pad.top + plotH - ((val - min) / range) * plotH;
ctx.strokeStyle = chartGrid;
ctx.lineWidth = 1;
for (let i = 0; i <= 4; i += 1) {
const y = pad.top + (plotH / 4) * i;
ctx.beginPath();
ctx.moveTo(pad.left, y);
ctx.lineTo(pad.left + plotW, y);
ctx.stroke();
}
ctx.beginPath();
ctx.moveTo(toX(0), toY(values[0]));
values.forEach((v, i) => ctx.lineTo(toX(i), toY(v)));
ctx.lineTo(toX(values.length - 1), pad.top + plotH);
ctx.lineTo(toX(0), pad.top + plotH);
ctx.closePath();
ctx.fillStyle = metric.fill;
ctx.fill();
ctx.beginPath();
values.forEach((v, i) => {
const x = toX(i);
const y = toY(v);
if (i === 0) ctx.moveTo(x, y);
else ctx.lineTo(x, y);
});
ctx.strokeStyle = metric.color;
ctx.lineWidth = 2.5;
ctx.stroke();
values.forEach((v, i) => {
const x = toX(i);
const y = toY(v);
ctx.beginPath();
ctx.arc(x, y, 4, 0, Math.PI * 2);
ctx.fillStyle = metric.color;
ctx.fill();
});
ctx.fillStyle = chartAxis;
ctx.font = '12px "IBM Plex Mono", monospace';
ctx.textAlign = "center";
rounds.forEach((r, i) => {
ctx.fillText(r, toX(i), h - 14);
});
ctx.save();
ctx.translate(16, h / 2);
ctx.rotate(-Math.PI / 2);
ctx.textAlign = "center";
const label = metric.labels[currentLang] || metric.labels.en;
ctx.fillText(label + (metric.unit ? ` (${metric.unit})` : ""), 0, 0);
ctx.restore();
ctx.textAlign = "right";
ctx.fillStyle = chartLatest;
const end = values[values.length - 1];
ctx.fillText(`${t("chartLatest")}: ${end}${metric.unit}`, w - 16, 18);
}
metricButtons.forEach((btn) => {
btn.addEventListener("click", () => {
metricButtons.forEach((item) => item.classList.remove("is-active"));
btn.classList.add("is-active");
activeMetric = btn.dataset.metric;
drawChart(activeMetric);
});
});
langButtons.forEach((btn) => {
btn.addEventListener("click", () => {
applyLanguage(btn.dataset.lang);
});
});
if (themeToggle) {
themeToggle.addEventListener("click", () => {
applyTheme(currentTheme === "dark" ? "light" : "dark");
});
}
window.addEventListener("resize", () => drawChart(activeMetric));
const revealElements = document.querySelectorAll(".reveal");
const observer = new IntersectionObserver(
(entries) => {
entries.forEach((entry) => {
if (entry.isIntersecting) {
entry.target.classList.add("in-view");
}
});
},
{ threshold: 0.12 }
);
revealElements.forEach((el) => observer.observe(el));
function animateCount(el, target, suffix = "") {
const duration = 900;
const start = performance.now();
const from = 0;
function tick(now) {
const progress = Math.min((now - start) / duration, 1);
const eased = 1 - Math.pow(1 - progress, 3);
const current = Math.round(from + (target - from) * eased);
el.textContent = `${current}${suffix}`;
if (progress < 1) requestAnimationFrame(tick);
}
requestAnimationFrame(tick);
}
function animateNumber(el, from, to, duration, formatter) {
const start = performance.now();
function tick(now) {
const progress = Math.min((now - start) / duration, 1);
const eased = 1 - Math.pow(1 - progress, 3);
const current = Math.round(from + (to - from) * eased);
el.textContent = formatter(current);
if (progress < 1) requestAnimationFrame(tick);
}
requestAnimationFrame(tick);
}
function parseNumericText(raw) {
const normalized = String(raw || "").replace(/[^\d]/g, "");
if (!normalized) return Number.NaN;
return Number(normalized);
}
function setupTrafficCounterAnimation(containerId, valueId) {
const activate = () => {
const valueEl = document.getElementById(valueId);
if (!valueEl) return false;
const target = parseNumericText(valueEl.textContent);
if (!Number.isFinite(target)) return false;
animateNumber(valueEl, 0, target, 1100, (value) => value.toLocaleString("en-US"));
return true;
};
if (activate()) return;
const containerEl = document.getElementById(containerId);
const observerTarget = containerEl || document.body;
const observer = new MutationObserver(() => {
if (activate()) observer.disconnect();
});
observer.observe(observerTarget, {
childList: true,
characterData: true,
subtree: true
});
setTimeout(() => observer.disconnect(), 30000);
}
const statValues = document.querySelectorAll(".stat-value[data-target]");
statValues.forEach((el) => {
const target = Number(el.dataset.target);
const suffix = el.dataset.suffix || "";
animateCount(el, target, suffix);
});
setupTrafficCounterAnimation("busuanzi_container_site_uv", "busuanzi_value_site_uv");
setupTrafficCounterAnimation("busuanzi_container_site_pv", "busuanzi_value_site_pv");
const copyButton = document.getElementById("copyBib");
const bibtex = document.getElementById("bibtex");
if (copyButton && bibtex) {
copyButton.addEventListener("click", () => {
navigator.clipboard
.writeText(bibtex.innerText.trim())
.then(() => {
copyButton.textContent = t("copied");
setTimeout(() => {
copyButton.textContent = t("copyBib");
}, 1200);
})
.catch(() => {
copyButton.textContent = t("copyFailed");
setTimeout(() => {
copyButton.textContent = t("copyBib");
}, 1200);
});
});
}
const yearEl = document.getElementById("year");
if (yearEl) {
yearEl.textContent = String(new Date().getFullYear());
}
applyTheme(currentTheme);
applyLanguage(currentLang);
drawChart(activeMetric);

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:7681c2076cba6a36046647861b03efb870aea38bba9b92b4d4db8cd916818e5e
size 26504161

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a38c59bc93ab9cb2f2d438debd98ba06bcb6d7a6aa8f7e092caa77591db2ffe7
size 21564949

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:e22f7cca684d15d66126bef75dbce9f3e8224d7eaa057133a8bbe151b8955002
size 15873071

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:798082de42937101e9320bce75859921ae7e26347c9a66f06a13be51888ce445
size 15879505

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:91ca17d0484ce7b91030c02f515af8da60291f7aa2fbd2649205c3b9e522a6d4
size 48327985

查看文件

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:fe94178b947be9a0d05e167cd82ebdb44ecf38b9376ca0588e99e49b3496dfc0
size 54471462

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 774 KiB

二进制
website/assets/images/evomind1.png 普通文件

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 28 KiB

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 89 KiB

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 160 KiB

二进制
website/assets/images/overview.png 普通文件

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 367 KiB

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 2.3 MiB

二进制
website/assets/images/sjtu.png 普通文件

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 297 KiB

二进制文件未显示。

之后

宽度:  |  高度:  |  大小: 481 KiB

查看文件

查看文件

@ -0,0 +1,12 @@
# Video Assets
Put your experiment videos here and keep these filenames so the webpage can render them directly:
- `value_overlay_demo.mp4`
- Suggested content: output from `outputs/value_infer/<RUN_NAME>/value/viz/` with value curve + ACP tag overlay.
- `policy_final_rollout.mp4`
- Suggested content: final policy execution on target task (best run).
- `intervention_compare.mp4`
- Suggested content: human intervention vs autonomous recovery comparison.
If a file is missing, the page still works and shows the poster image.

373
website/index.html 普通文件
查看文件

@ -0,0 +1,373 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Evo-RL | Continuous Open-Source Real-World RL</title>
<meta name="description" content="Evo-RL: continuous open-source real-world reinforcement learning on SO101 and AgileX PiPER, with a full reproducible CLI pipeline." />
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link href="https://fonts.googleapis.com/css2?family=Chakra+Petch:wght@400;500;600;700&family=Newsreader:opsz,wght@6..72,400;6..72,500;6..72,700&family=IBM+Plex+Mono:wght@400;500&display=swap" rel="stylesheet" />
<link rel="stylesheet" href="styles.css" />
</head>
<body>
<div class="bg-layer"></div>
<div class="grid-layer"></div>
<header class="site-header">
<a class="brand" href="#top">
<span class="brand-mark">EVO</span>
<span class="brand-name">Evo-RL</span>
</a>
<div class="header-right">
<nav class="site-nav">
<a href="#focus">Focus</a>
<a href="#pipeline">Pipeline</a>
<a href="#results">Results</a>
<a href="#media">Media</a>
<a href="#reproduce">Reproduce</a>
</nav>
<div class="site-controls">
<div class="lang-switch" role="group" aria-label="Language switch">
<button id="langEn" class="lang-btn is-active" type="button" data-lang="en">EN</button>
<button id="langZh" class="lang-btn" type="button" data-lang="zh">中</button>
</div>
<button id="themeToggle" class="theme-toggle" type="button" aria-label="Theme toggle">
<span class="theme-icon" id="themeIcon">◐</span>
<span id="themeLabel">Night</span>
</button>
</div>
</div>
</header>
<main id="top">
<section id="traffic" class="traffic-strip">
<div class="traffic-widget">
<p class="traffic-title">Site Traffic</p>
<div class="traffic-item">
<span class="traffic-label">Visitors</span>
<strong id="busuanzi_container_site_uv"><span id="busuanzi_value_site_uv">--</span></strong>
</div>
<div class="traffic-item">
<span class="traffic-label">Views</span>
<strong id="busuanzi_container_site_pv"><span id="busuanzi_value_site_pv">--</span></strong>
</div>
</div>
</section>
<section class="hero section reveal">
<div class="hero-left">
<p class="eyebrow">Continuous Open-Source Real-World RL</p>
<h1>
Evo-RL
<span>SO101 to AgileX PiPER</span>
</h1>
<p class="hero-summary">
A LeRobot-aligned, full-pipeline real-world offline RL workflow for real robots: collect rollout data,
train value functions and policies offline, infer advantage tags, and iterate deployment in a closed loop.
</p>
<div class="news-pill">
<span>NEWS</span>
<strong>February 26, 2026</strong>
<em>First SO101 real-world RL baseline + reproducible CLI released.</em>
</div>
<div class="hero-actions">
<a class="btn btn-primary" href="https://github.com/MINT-SJTU/Evo-RL" target="_blank" rel="noreferrer">GitHub Repo</a>
<a class="btn btn-ghost" href="https://huggingface.co/docs/lerobot/index" target="_blank" rel="noreferrer">LeRobot Docs</a>
</div>
</div>
<div class="hero-right">
<div class="hero-card shell-card">
<p class="shell-title">pipeline.sh</p>
<pre><code>lerobot-human-inloop-record
--dataset.repo_id=&lt;DATASET&gt;
lerobot-value-train
--value.type=pistar06
lerobot-value-infer
--acp.positive_ratio=0.3
lerobot-train
--acp.enable=true</code></pre>
</div>
<div class="hero-stats">
<article class="stat-card">
<p class="stat-label">Robot Platforms</p>
<p class="stat-value" data-target="2">0</p>
</article>
<article class="stat-card">
<p class="stat-label">Open Task Datasets</p>
<p class="stat-value" data-target="0">0</p>
</article>
<article class="stat-card">
<p class="stat-label">Open Models</p>
<p class="stat-value" data-target="0">0</p>
</article>
<article class="stat-card">
<p class="stat-label">Release Year</p>
<p class="stat-value" data-target="2026">0</p>
</article>
</div>
</div>
</section>
<section class="logo-row section reveal">
<img src="assets/images/sjtu.png" alt="SJTU community visual" />
<img src="assets/images/evomind1.png" alt="EvoMind" />
</section>
<section id="focus" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Project Focus</p>
<h2>Open Infrastructure for Real-World RL</h2>
</div>
<div class="card-grid three">
<article class="glass-card">
<h3>Open real-world RL pipeline</h3>
<p>Runnable workflows on SO101 and AgileX PiPER covering collection, offline value/policy training, and iterative deployment.</p>
</article>
<article class="glass-card">
<h3>Reproducible assets</h3>
<p>Codebase, model layout, dataset structure, and command-line templates are released to reduce setup friction and improve reproducibility.</p>
</article>
<article class="glass-card">
<h3>Community co-evolution</h3>
<p>Reproduce existing real-world RL methods, extend value/policy variants, and continuously publish experiments and benchmarks for collective progress.</p>
</article>
</div>
</section>
<section id="pipeline" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Method</p>
<h2>6-Stage Closed-Loop Pipeline</h2>
</div>
<ol class="pipeline-list">
<li><span>01</span><p><strong>System preparation</strong> to ensure robot, sensors, and control stack are ready.</p></li>
<li><span>02</span><p><strong>Interactive data collection</strong> with human guidance and episode-level outcomes.</p></li>
<li><span>03</span><p><strong>Value modeling</strong> to estimate trajectory quality from accumulated data.</p></li>
<li><span>04</span><p><strong>Signal annotation</strong> to convert value outputs into usable training supervision.</p></li>
<li><span>05</span><p><strong>Policy update and deployment</strong> to run the improved policy in real tasks.</p></li>
<li><span>06</span><p><strong>Data refresh and iteration</strong> to continue improving robustness and success rate.</p></li>
</ol>
</section>
<section id="results" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Results</p>
<h2>Training Signals Across Iterations</h2>
</div>
<div class="results-layout">
<article class="glass-card chart-card">
<div class="chart-toolbar">
<button class="metric-btn is-active" data-metric="value">Value</button>
<button class="metric-btn" data-metric="advantage">Advantage</button>
<button class="metric-btn" data-metric="success">Success Rate</button>
</div>
<canvas id="signalChart" width="960" height="480" aria-label="Evo-RL signals chart"></canvas>
<p class="chart-note">Illustrative trend view for project webpage. Replace with your run logs for paper-ready plots.</p>
</article>
<article class="glass-card insights-card">
<h3>What This Visual Captures</h3>
<ul>
<li>Value trend stabilizes as rollouts and labels improve across rounds.</li>
<li>Advantage separates better trajectories used for ACP tagging.</li>
<li>Success rate rises with iterative deployment and data refresh.</li>
</ul>
<p class="note-strong">Current default ACP ratio in README pipeline: <code>--acp.positive_ratio=0.3</code>.</p>
</article>
</div>
</section>
<section id="media" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Qualitative Demos</p>
<h2>Visual Results Gallery</h2>
</div>
<div class="viz-board">
<article class="glass-card viz-lane viz-lane-value">
<header class="viz-head">
<p class="lane-kicker">VALUE VISUAL</p>
<h3>Success and Failure Case</h3>
<p class="lane-note">Frame-level value visualization on real robot rollouts.</p>
</header>
<div class="viz-strip viz-strip-stack">
<figure class="viz-cell">
<img src="assets/gifs/value_success.gif" alt="Value visual success case" loading="lazy" />
</figure>
<figure class="viz-cell">
<img src="assets/gifs/value_failure.gif" alt="Value visual failure case" loading="lazy" />
</figure>
</div>
</article>
<article class="glass-card viz-lane">
<header class="viz-head">
<p class="lane-kicker">POLICY ROLLOUT</p>
<h3>Deployment Visual Results</h3>
<p class="lane-note">Final policy behavior clips in target task execution.</p>
</header>
<div class="viz-strip">
<figure class="viz-cell">
<img src="assets/gifs/policy_rollout_1.gif" alt="Policy rollout result 1" loading="lazy" />
</figure>
<figure class="viz-cell">
<img src="assets/gifs/policy_rollout_2.gif" alt="Policy rollout result 2" loading="lazy" />
</figure>
</div>
</article>
<article class="glass-card viz-lane">
<header class="viz-head">
<p class="lane-kicker">HUMAN-IN-THE-LOOP</p>
<h3>Intervention and Recovery Results</h3>
<p class="lane-note">Human guidance segments and autonomous recovery outcomes.</p>
</header>
<div class="viz-strip">
<figure class="viz-cell">
<img src="assets/gifs/hitl_1.gif" alt="Human-in-the-loop result 1" loading="lazy" />
</figure>
<figure class="viz-cell">
<img src="assets/gifs/hitl_2.gif" alt="Human-in-the-loop result 2" loading="lazy" />
</figure>
</div>
</article>
</div>
</section>
<section id="architecture" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Architecture</p>
<h2>LeRobot-Aligned VLA + Value Loop</h2>
</div>
<div class="architecture-block glass-card">
<img src="assets/images/VLA_architecture.jpg" alt="VLA architecture" />
<div>
<p>
Evo-RL reuses LeRobot's practical data and inference stack while adding value-guided
filtering signals (`value`, `advantage`, `acp_indicator`) to improve iterative policy training.
</p>
<p>
The same design supports single-GPU and accelerate-based multi-GPU workflows for both value and policy stages.
</p>
</div>
</div>
</section>
<section id="related" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Related Work</p>
<h2>Context in Open Robot Learning</h2>
</div>
<article class="glass-card link-cloud">
<a href="https://www.pi.website/blog/pistar06" target="_blank" rel="noreferrer">Pi*0.6</a>
<a href="https://www.314159.com/blog" target="_blank" rel="noreferrer">Pi0.5</a>
<a href="https://github.com/MINT-SJTU/Evo-1" target="_blank" rel="noreferrer">EVO-1</a>
<a href="https://github.com/MINT-SJTU/Evo-SOTA.io" target="_blank" rel="noreferrer">EVO-SOTA</a>
<a href="https://github.com/huggingface/lerobot" target="_blank" rel="noreferrer">LeRobot</a>
</article>
</section>
<section id="reproduce" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Reproducibility</p>
<h2>Core CLI Commands</h2>
</div>
<div class="card-grid three">
<article class="glass-card">
<h3>Human-in-loop Record</h3>
<pre><code>lerobot-human-inloop-record \
--robot.type=&lt;ROBOT_TYPE&gt; \
--teleop.type=&lt;TELEOP_TYPE&gt; \
--dataset.repo_id=&lt;HF_USER_OR_ORG/DATASET&gt; \
--dataset.single_task="&lt;TASK_TEXT&gt;" \
--dataset.num_episodes=&lt;NUM_EPISODES&gt; \
--display_data=true</code></pre>
</article>
<article class="glass-card">
<h3>Value Training</h3>
<pre><code>lerobot-value-train \
--dataset.repo_id=&lt;HF_USER_OR_ORG/DATASET&gt; \
--value.type=&lt;VALUE_TYPE&gt; \
--batch_size=&lt;BATCH_SIZE&gt; \
--output_dir=outputs/value_train/&lt;RUN_NAME&gt; \
--job_name=&lt;RUN_NAME&gt;</code></pre>
</article>
<article class="glass-card">
<h3>Policy Training</h3>
<pre><code>lerobot-train \
--dataset.repo_id=&lt;HF_USER_OR_ORG/DATASET&gt; \
--policy.type=&lt;POLICY_TYPE&gt; \
--steps=&lt;TRAIN_STEPS&gt; \
--acp.enable=true \
--acp.indicator_dropout_prob=0.3 \
--output_dir=outputs/train/&lt;RUN_NAME&gt;</code></pre>
</article>
</div>
</section>
<section id="community" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Model & Community</p>
<h2>Links and Contact</h2>
</div>
<div class="card-grid two community-grid">
<article class="glass-card">
<h3>Release Status</h3>
<ul>
<li>Hugging Face model: coming soon</li>
<li>Hugging Face dataset: coming soon</li>
<li>Canonical version tags will be pinned once published</li>
</ul>
</article>
<article class="glass-card qr-card">
<h3>WeChat Group</h3>
<p class="qr-desc">Scan to join EvoMind's WeChat community and follow project updates.</p>
<div class="qr-frame">
<img src="assets/images/evomind_wechat.jpg" alt="EvoMind WeChat QR code" />
</div>
<p class="qr-caption">
<span class="qr-contact-label">Contact</span>
<a class="qr-email" href="mailto:business@evomind-tech.com">business@evomind-tech.com</a>
</p>
</article>
</div>
</section>
<section id="citation" class="section reveal">
<div class="section-heading">
<p class="eyebrow">Citation</p>
<h2>BibTeX</h2>
</div>
<article class="glass-card citation-card">
<button class="copy-btn" id="copyBib" type="button">Copy BibTeX</button>
<pre id="bibtex"><code>@misc{evorl2026,
title = {Evo-RL: Continuous Open-Source Real-World RL on SO101 and Beyond},
author = {Evo-RL Contributors},
year = {2026},
howpublished = {\url{https://github.com/MINT-SJTU/Evo-RL}}
}</code></pre>
</article>
</section>
</main>
<footer class="site-footer">
<p>Evo-RL • Apache-2.0 • <span id="year"></span></p>
</footer>
<script async src="https://busuanzi.ibruce.info/busuanzi/2.3/busuanzi.pure.mini.js"></script>
<script src="app.js"></script>
</body>
</html>

1011
website/styles.css 普通文件

文件差异内容过多而无法显示 加载差异