# Quiz

1. What is the advantage of using simulation in robot learning?
   - **A.** It’s more cost-effective
   - **B.** It allows for faster than real-time training
   - **C.** It provides access to privileged information
   - **D.** All of the above

### Answer

**D**

1. The “reality gap” in sim-to-real transfer refers to:
   - **A.** The difference between simulation and real-world conditions
   - **B.** The time it takes to deploy a policy
   - **C.** The cost difference between simulation and real robots
   - **D.** The physical distance between the simulation computer and the robot

### Answer

**A**

1. Which of the following is not a component of the reality gap?
   - **A.** Approximation errors
   - **B.** Model errors
   - **C.** Unmodelled dynamics
   - **D.** User interface errors

### Answer

**D**

1. Domain randomization always improves the performance of a policy when deployed on hardware.
   - **A.** True
   - **B.** False

### Answer

**B, false**

1. What is the primary purpose of regularization in sim-to-real transfer?
   - **A.** To increase training speed
   - **B.** To create more realistic simulations
   - **C.** To encourage smooth, hardware-friendly behavior
   - **D.** To improve visual rendering in simulations

### Answer

**C**

1. In the context of privileged information, what does the “teacher” in a teacher-student approach have access to?
   - **A.** Only real-world data
   - **B.** Only simulated data
   - **C.** Both simulated and real-world data
   - **D.** Information not available to the real robot

### Answer

**D**

1. Which of the following is a benefit of using actuator networks in sim-to-real transfer?
   - **A.** They eliminate the need for real robots
   - **B.** They accurately model complex actuator behaviors
   - **C.** They speed up physical robot movement
   - **D.** They reduce the cost of robot hardware

### Answer

**B**

1. The goal of visual domain randomization is to achieve photorealism.
   - **A.** True
   - **B.** False

### Answer

**B, False**

1. What is a key consideration when balancing domain randomization and real-to-sim approaches?
   - **A.** Always prioritize domain randomization
   - **B.** Focus solely on real-to-sim techniques
   - **C.** Find a balance that creates robust yet effective policies
   - **D.** Ignore both approaches and rely on intuition

### Answer

**C**