threejs 实现一个简易的泊车

在上一篇文章 threejs都有些啥搭建的场景的基础上，咱们尝试做一个简单的自动泊车，进一步探究下 threejs 中还有哪些东西

简易版小车

因为之前用的模型比较大，加载很慢，这里就先自己简单实现一辆小车（后面统称自车），如下图：

将车轮、车体和边框组合成一个 Group，便于后面做自车的一些操作，实现代码如下：

ts 复制代码

 // 自车车体
const geometry = new THREE.BoxGeometry(2, 0.6, 3);
const material = new THREE.MeshBasicMaterial({
  color: 0x00ffff,
  side: THREE.DoubleSide,
});
const vehicle = new THREE.Mesh(geometry, material);
vehicle.position.set(0, 1, 0);
scene.add(vehicle);
// 增加自车边框
const box = geometry.clone();
const edges = new THREE.EdgesGeometry(box);
const edgesMaterial = new THREE.LineBasicMaterial({
  color: 0x333333,
});
const line = new THREE.LineSegments(edges, edgesMaterial);
line.position.x = 0;
line.position.y = 1;
line.position.z = 0;
scene.add(line);
// 组成一个Group
const egoCar = new THREE.Group();
egoCar.name = "自车";
egoCar.add(vehicle, line);
scene.add(egoCar);
// 车轮
const axlewidth = 0.7;
const radius = 0.4;
const wheels: any[] = [];
const wheelObjects: any[] = [];
wheels.push({ position: [axlewidth, 0.4, -1], radius });
wheels.push({
  position: [-axlewidth, 0.4, -1],
  radius,
});
wheels.push({ position: [axlewidth, 0.4, 1], radius });
wheels.push({ position: [-axlewidth, 0.4, 1], radius });
wheels.forEach(function (wheel) {
  const geometry = new THREE.CylinderGeometry(
    wheel.radius,
    wheel.radius,
    0.4,
    32
  );
  const material = new THREE.MeshPhongMaterial({
    color: 0xd0901d,
    emissive: 0xee0000,
    side: THREE.DoubleSide,
    flatShading: true,
  });
  const cylinder = new THREE.Mesh(geometry, material);
  cylinder.geometry.rotateZ(Math.PI / 2);
  cylinder.position.set(
    wheel.position[0],
    wheel.position[1],
    wheel.position[2]
  );
  egoCar.add(cylinder);
  // 后面修改车轮方向会用到
  wheelObjects.push(cylinder);
});

跟车相机

让相机一直跟着自车，体验更好一点

ts 复制代码

// ...
const camera = new THREE.PerspectiveCamera(45, width / height, 0.1, 800);
// 设置摄像机位置，并将其朝向场景中心
camera.position.x = 0;
// camera.position.y = 10;
// camera.position.z = 20;
// camera.lookAt(scene.position);
camera.lookAt(egoCar.position);
// ...
function animate() {
  stats.begin();
  controls.update();
  // 相机跟随自车
  camera.position.y = egoCar.position.y + 15;
  camera.position.z = egoCar.position.z + 25;
  camera.lookAt(egoCar.position);
  renderer.render(scene, camera);
  stats.end();
  requestAnimationFrame(animate);
}
// ...

自车行驶

实现自车前行后退和左右转向

ts 复制代码

// ...
// 记录开始按下的时间
let startTime = 0;
const activeKeys = new Set();
let t = 0;
document.addEventListener("keydown", (e) => {
  activeKeys.add(e.key);
  if (startTime === 0) {
    startTime = Date.now();
  }
  t = (Date.now() - startTime) / 1000;
  if (t > 10) {
    t = 10;
  }
});
document.addEventListener("keyup", (e) => {
  activeKeys.delete(e.key);
  if (activeKeys.size === 0) {
    startTime = 0;
  }
});
function animate() {
  stats.begin();
  controls.update();
  // 相机跟随自车
  camera.position.y = egoCar.position.y + 15;
  camera.position.z = egoCar.position.z + 25;
  camera.lookAt(egoCar.position);
  if (activeKeys.has("ArrowUp")) {
    // 估算对应方向的移动距离
    egoCar.position.z -= t * 0.1 * Math.cos(egoCar.rotation.y);
    egoCar.position.x -= t * 0.1 * Math.sin(egoCar.rotation.y);
  }
  if (activeKeys.has("ArrowDown")) {
    egoCar.position.z += t * 0.1 * Math.cos(egoCar.rotation.y);
    egoCar.position.x += t * 0.1 * Math.sin(egoCar.rotation.y);
  }
  if (activeKeys.has("ArrowLeft")) {
    egoCar.rotation.y += 0.01;
  }
  if (activeKeys.has("ArrowRight")) {
    egoCar.rotation.y -= 0.01;
  }
  renderer.render(scene, camera);
  stats.end();
  requestAnimationFrame(animate);
}
//...

车轮转动

遍历车轮对象，动态修改车轮的偏转角 rotation，以车头方向为基准偏转固定的角度

ts 复制代码

function animate() {
  // ...
  if (activeKeys.has("ArrowLeft")) {
    egoCar.rotation.y += 0.01;
    wheelObjects.forEach((wheel) => {
      wheel.rotation.y = egoCar.rotation.y + Math.PI / 4;
    });
  }
  if (activeKeys.has("ArrowRight")) {
    egoCar.rotation.y -= 0.01;
    wheelObjects.forEach((wheel) => {
      wheel.rotation.y = egoCar.rotation.y - Math.PI / 4;
    });
  }
  // ...
}

行进效果还是有点僵硬(能用就行)，这里的问题是行进方向应该是按车头方向，而不是固定按某个坐标轴方向，不过这里也只是简单模拟这个行进效果，后面再引入物理库 cannon.js优化下这块控制逻辑

泊车功能

车位实现

做一个贴地面的矩形框来模拟车位，可以使用 THREE.PlaneGeometry 来创建平面几何体

ts 复制代码

createParkingSpace() {
    const plane = new THREE.PlaneGeometry(8, 5);
    const material = new THREE.MeshPhongMaterial({
      color: 0x666666,
      side: THREE.DoubleSide,
    });
    const mesh = new THREE.Mesh(plane, material);
    mesh.rotation.x = -Math.PI / 2;
    mesh.position.set(10, 0.12, -20);
    this.scene?.add(mesh);
    // 增加自定义type，便于后面处理车位的选中逻辑
    mesh.userData.type = "parkingSpace";
}

现在咱们把小车开过去停到那个位置

自动泊车

需要实现点击车位后高亮对应的车位，之后小车自动行驶到对应的位置并停好。点击原理是用射线的方式采集第一个碰到的车位物体，当点击鼠标时，会发生以下步骤：

基于屏幕上的点击位置创建一个 THREE.Vector3 向量
使用 vector.unproject 方法将屏幕上点击位置的坐标转换成 three.js 场景中的坐标
创建 THREE.Raycaster可以从摄像机的位置向场景中鼠标的点击位置发出一条射线
raycaster.intersectObjects 返回包含了所有被射线穿过的对象信息的数组(从摄像机位置开始由短到长)

ts 复制代码

function handleParkSpaceClick(event: any) {
  let vector = new THREE.Vector3(
    (event.clientX / window.innerWidth) * 2 - 1,
    -(event.clientY / window.innerHeight) * 2 + 1,
    0.5
  );
  vector = vector.unproject(camera);
  const raycaster = new THREE.Raycaster(
    camera.position,
    vector.sub(camera.position).normalize()
  );
  const intersects = raycaster.intersectObjects(scene.children);
  for (let i = 0; i < intersects.length; i++) {
    const obj = intersects[i];
    // @ts-ignore
    if (obj.object.userData.type === "parkingSpace")
      // @ts-ignore
      obj.object.material.color.set(0x00ff00);
  }
}
document.addEventListener("click", handleParkSpaceClick);

自动泊车的实现逻辑也比较简单，这里简单记住了车位的位置信息，然后让小车按一定的偏移驶入，其实实际场景可能还要考虑躲避障碍物、加减速、偏转角等，一般也不由前端操心这些。实现代码参考 three-gta v0.1.1 -- 在线体验