2b:null 2c:["$","title",null,{"children":"tfjs inverse kinematics 逆运动学体验 | Bruce He | Observable"}] 31:T876,{ const [motor0, motor1, motor2, target] = nodes; const dom = html` `; const vMotor0 = dom.querySelector("#motor0"); const vMotor1 = dom.querySelector("#motor1"); const vMotor2 = dom.querySelector("#motor2"); const vLine0 = dom.querySelector("#line0"); const vLine1 = dom.querySelector("#line1"); const vTarget = dom.querySelector("#target"); drag(target)(d3.select(vTarget)); //motor1.fx = null; //motor1.fx = null; while (true) { let [newM1Pos, newM2Pos] = updatePos(target); motor1.fx = newM1Pos[0]; motor1.fy = newM1Pos[1]; motor2.fx = newM2Pos[0]; motor2.fy = newM2Pos[1]; vTarget.setAttribute("cx", target.fx); vTarget.setAttribute("cy", target.fy); vMotor0.setAttribute("cx", motor0.fx); vMotor0.setAttribute("cy", motor0.fy); vMotor1.setAttribute("cx", motor1.fx); vMotor1.setAttribute("cy", motor1.fy); vMotor2.setAttribute("cx", motor2.fx); vMotor2.setAttribute("cy", motor2.fy); vLine0.setAttribute("x2", motor1.fx); vLine0.setAttribute("y2", motor1.fy); vLine1.setAttribute("x1", motor1.fx); vLine1.setAttribute("y1", motor1.fy); vLine1.setAttribute("x2", motor2.fx); vLine1.setAttribute("y2", motor2.fy); //yield Promises.delay(3000); yield dom; } }32:[] 2f:{"id":"7c8f11f718614c36","slug":"tfjs-inverse-kinematics","title":"tfjs inverse kinematics 逆运动学体验","thumbnail":"1d7d2440f2291e78fafbb5ab525a93211aac72d4c96e7bb6e267262338a6b06e","access_level":"public","create_time":"$D2021-03-12T02:18:25.816Z","update_time":"$D2022-05-15T12:36:40.053Z","view_time":null,"first_public_time":"$D2022-03-27T06:13:17.800Z","fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"5e3282f592b2837f","login":"heganjie","name":"Bruce He","avatar_url":"https://avatars-next.observableusercontent.com/58d33228761b95f76b498c59d4664893e0c8b694acb62ff0342f8d22ebafb8e3","tier":"starter_2024"},"version":368,"publish_version":368,"role":null,"authored":false,"authors":[{"type":"individual","id":"8bc3b438d4504c1b","login":"heganjie","name":"Bruce He","avatar_url":"https://avatars-next.observableusercontent.com/58d33228761b95f76b498c59d4664893e0c8b694acb62ff0342f8d22ebafb8e3","tier":null}],"members":[{"type":"individual","id":"8bc3b438d4504c1b","login":"heganjie","name":"Bruce He","avatar_url":"https://avatars-next.observableusercontent.com/58d33228761b95f76b498c59d4664893e0c8b694acb62ff0342f8d22ebafb8e3","tier":null,"role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"subscribed":null,"forks":0,"collection":{"id":"40d740bf395b0cdf","slug":"untitled","title":"algorithm","owner":{"id":"5e3282f592b2837f","login":"heganjie","name":"Bruce He"}},"type":"notebook","body":{"title":"tfjs inverse kinematics 逆运动学体验","cells":[{"id":0,"value":"# tfjs inverse kinematics 逆运动学体验\n\n【GAMES101-现代计算机图形学入门-闫令琪-哔哩哔哩】【视频标记点 1:02:47】 https://b23.tv/9lwpfNl\n\nhttps://observablehq.com/@heganjie/d3-force-demo\n\nhttps://observablehq.com/@bumbeishvili/tfjs-fitting-a-curve-to-synthetic-data\n\nhttps://www.tensorflow.org/js/guide/tensors_operations\n\nhttps://github.com/tensorflow/tfjs\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined"},{"id":13,"value":"pinRadius = 5","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":22,"value":"armLen = 40","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":12,"value":"nodes = {\n let m0 = { id: \"motor0\", fx: 200 - pinRadius, fy: 100, r: pinRadius };\n let m1 = { id: \"motor1\", fx: m0.fx + armLen, fy: 100, r: pinRadius };\n return [\n m0,\n m1,\n { id: \"motor2\", fx: m1.fx + armLen, fy: 100, r: pinRadius },\n { id: \"target\", fx: 200, fy: 25, r: pinRadius }\n ];\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":31,"value":"$31","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":81,"value":"# 单轴\n${tex`p_x=l_1cos(\\theta_1)`}，${tex`p_y=l_1sin(\\theta_1)`}\n\n# 双轴公式\n${tex`p_x=l_1cos(\\theta_1)+l_2cos(\\theta_1+\\theta_2)`}，${tex`p_y=l_1sin(\\theta_1)+l_2sin(\\theta_1+\\theta_2)`}\n\n# 损失函数\n${tex`J(\\theta_1)=0.25*((O_x-P_x)^2+(O_y-P_y)^2)`}\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined"},{"id":147,"value":"t1 = tf.variable(tf.scalar(Math.random()))","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":344,"value":"t2 = tf.variable(tf.scalar(Math.random()))","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":199,"value":"makeRotMat = theta => {\n let cosPart = tf.cos(theta).mul(\n tf.tensor2d([\n [1, 0],\n [0, 1]\n ])\n );\n let sinPart = tf.sin(theta).mul(\n tf.tensor2d([\n [0, -1],\n [1, 0]\n ])\n );\n return cosPart.add(sinPart); //t1Rot.dataSync()\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":280,"value":"m0Pos = {\n let [m0] = nodes;\n return tf.tensor2d([m0.fx, m0.fy], [1, 2]);\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":277,"value":"vec0 = tf.tensor2d([armLen, 0], [1, 2])","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":144,"value":"function predict(startPos, theta) {\n return tf.tidy(() => {\n let vecRot = tf.matMul(vec0, makeRotMat(theta));\n return startPos.add(vecRot);\n });\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":149,"value":"function distance(p1, p2) {\n return p1.sub(p2).square().sum().sqrt();\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":155,"value":"Optimizer = tf.train.adam(0.06)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":32,"value":"updatePos = (target) => {\n // let [m0, m1, target] = nodes;\n // 初始值 t1, 输入 px, py，输出新的 t1\n // update m1 pos\n\n // const predsYs = predict(m0Pos);\n // return distance(predsYs, pPos).dataSync();\n const pPos = tf.tensor2d([target.fx, target.fy], [1, 2]);\n Optimizer.minimize(() => {\n const m1Pos = predict(m0Pos, t1);\n const m2Pos = predict(m1Pos, t2);\n return distance(m2Pos, pPos);\n });\n\n let newM1Pos = predict(m0Pos, t1);\n let newM2Pos = predict(newM1Pos, t2);\n return [newM1Pos.dataSync(), newM2Pos.dataSync()];\n}","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":28,"value":"drag = (d) => {\n function dragstarted(event) {\n //if (!event.active) simulation.alphaTarget(0.3).restart();\n //d.fx = d.x;\n //d.fy = d.y;\n }\n\n function dragged(event) {\n d.fx = event.x;\n d.fy = event.y;\n }\n\n function dragended(event) {\n //if (!event.active) simulation.alphaTarget(0);\n //d.fx = null;\n //d.fy = null;\n }\n\n return d3\n .drag()\n .on(\"start\", dragstarted)\n .on(\"drag\", dragged)\n .on(\"end\", dragended);\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":61,"value":"tf = require(\"@tensorflow/tfjs\")\n//tf = require(\"https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@3.14.0/dist/tf.js\")","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"},{"id":27,"value":"d3 = require(\"d3\");","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined"}],"files":"$W32","stdlib":"1","titling":"auto"}} 30:"$2f:body"