Physical Intelligence and pi-zero Quiz

Answers and explanations appear as you take the quiz.

1. What type of model is pi-zero?
   • A vision-language-action (VLA) model
   • A text-only chatbot
   • An image generator
   • A speech recognition model
2. Which manipulation task did the pi-zero paper demonstrate after task-specific training?
   • Mowing a lawn
   • Folding laundry
   • Washing a car
   • Painting a wall
3. Which technique does pi-zero use to generate smooth continuous robot actions?
   • Hard-coded if-then rules
   • A simple lookup table
   • Flow matching (a diffusion-like method)
   • GPS waypoint following
4. What backbone does pi-zero build upon to understand instructions and scenes?
   • A physics simulator only
   • A relational database
   • A spreadsheet engine
   • A pretrained vision-language model
5. How many distinct robot configurations are described in pi-zero’s internal training data?
   • Eight
   • One
   • Two
   • Thirty-two
6. What is Physical Intelligence's core strategic goal?
   • Selling one bespoke skill per robot
   • Building a general foundation model that works across many robots
   • Manufacturing only humanoid bodies
   • Building self-driving cars
7. Which open-source robot dataset is included in pi-zero’s training mixture?
   • ImageNet
   • Common Voice
   • Open X-Embodiment
   • MUSDB18
8. At what frequency does the pi-zero paper say the model can output motor commands for dexterous control?
   • Once per minute
   • One frame per hour
   • Only after a human approves each action
   • Up to 50 times per second