引导式测验 · 中等 等级
特斯拉人工智能和自动驾驶仪测验
回顾一下特斯拉基于摄像头的感知和驾驶辅助系统，包括对专心的人类驾驶员的持续需求。
问题 1 的 8
In Tesla’s camera-based “Tesla Vision” approach, what sensors provide the visual input?
预览每个问题8 问题
当您参加测验时，就会出现答案和解释。
- In Tesla’s camera-based “Tesla Vision” approach, what sensors provide the visual input?
- Cameras around the vehicle
- A lidar-only scanning unit
- Radar without cameras
- A single ultrasonic sensor
- 尽管名称如此，特斯拉的“完全自动驾驶（监督）”目前需要什么？
- 根本没有司机
- 细心的人类司机准备接管
- 第二辆车跟在后面
- 控制中心的远程操作员
- What do Tesla’s per-camera neural networks analyze in the company’s AI description?
- Only spoken commands
- Satellite maps without camera data
- Raw images for semantic segmentation, object detection and monocular depth estimation
- Vehicle registration documents
- 特斯拉从其庞大的车队中获得了什么优势？
- 培训免费用电
- 自动政府批准
- 更便宜的轮胎
- 用于训练数据的大量真实驾驶视频
- Which humanoid robot project appears on Tesla’s AI and Robotics page?
- 擎天柱
- 阿特拉斯
- Digit
- 现货
- What do Tesla’s birds-eye-view networks output from video across vehicle cameras?
- Only the car’s color
- Road layout, static infrastructure and 3D objects in a top-down view
- A transcript of nearby speech
- A single two-dimensional lane label
- How does Tesla say it creates ground-truth data for training networks to predict world representations?
- By using only written driver comments
- By copying another company’s labels
- By combining sensor information across space and time
- By asking the model to label itself without sensor data
- What evaluation methods does Tesla say it uses to track improvements and prevent regressions?
- Only user ratings in an app store
- A single annual road test
- Only language-model benchmarks
- Open- and closed-loop plus hardware-in-the-loop evaluation infrastructure