Ikibazo gihuza ibibazo · Hagati Urwego

Tesla AI hamwe na Autopilot Ikibazo

Ongera usuzume kamera ya Tesla yerekana kamera hamwe na sisitemu yo gufasha abashoferi, harimo gukomeza gukenera umushoferi wabantu witonze.

Inzira zijyanye no kuyoboraTesla AI na Autopilot
Ikibazo 1 Bya 8

In Tesla’s camera-based “Tesla Vision” approach, what sensors provide the visual input?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. 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
  2. Nubwo izina ryayo, 'Tesla' Yuzuye-Gutwara (Kugenzurwa) 'bisaba iki muri iki gihe?
    • Nta mushoferi na gato
    • Umushoferi wumuntu witonze yiteguye gufata
    • Imodoka ya kabiri ikurikira inyuma
    • Umukoresha wa kure mugenzuzi
  3. 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
  4. Ni izihe nyungu Tesla yunguka mumato manini manini mumuhanda?
    • Amashanyarazi yubusa yo guhugura
    • Icyemezo cya leta cyikora
    • Amapine ahendutse
    • Umubare munini wa videwo nyayo yo gutwara ibinyabiziga
  5. Which humanoid robot project appears on Tesla’s AI and Robotics page?
    • Optimus
    • Atlas
    • Digit
    • Ikibanza
  6. 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
  7. 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
  8. 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