Testing Autonomous AI-Based Systems CT-AI Practice Questions (Page 5)

Questions 21–25

1. 21application · medium

   A company is testing an autonomous agricultural tractor that operates at SAE Level 2 autonomy: it can steer and control speed automatically, but a human operator must remain attentive and ready to take over at any time. The tractor's path-planning algorithm is being updated. The team wants to test the new algorithm's behavior in a field with unexpected obstacles. What is the most important testing consideration given the tractor's autonomy level?

   • AThe test must focus on the human-machine interface, ensuring the operator can effectively monitor the system and take over when the tractor encounters an obstacle it cannot handle.
   • BThe test must be conducted entirely in simulation because real-world testing is not possible for Level 2 systems.
   • CThe test must verify that the tractor can navigate the field without the operator needing to pay attention, since the system handles all driving tasks.
   • DThe test must verify that the tractor can handle all obstacles without any human intervention, since the system is fully autonomous.
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   Which of the following is an example of environmental variability that complicates testing of autonomous systems?

   • AChanges in the system's memory allocation patterns
   • BChanges in lighting conditions affecting a self-driving car's perception system
   • CDifferences in the programming language used to implement the system
   • DVariations in the system's internal clock speed
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   A team is testing an autonomous drone used for agricultural monitoring. The drone's flight controller is deterministic, but its obstacle avoidance system uses a neural network that is non-deterministic. The team has observed that the drone occasionally takes slightly different paths to avoid the same obstacle in repeated tests. The team needs to determine whether this variability is acceptable. What is the most appropriate approach?

   • ARun the same test multiple times and analyze the distribution of paths taken to determine if all paths are safe and within acceptable bounds.
   • BUse formal verification to prove that the neural network will always produce the same path for the same obstacle.
   • CTreat the variability as a defect and fix the neural network to produce deterministic outputs.
   • DIgnore the variability because the drone always avoids the obstacle, regardless of the path taken.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A team is testing an autonomous lawn mower that operates in residential yards. The mower uses a camera to detect obstacles and a GPS to navigate. During field testing, the team notices that the mower behaves differently on cloudy days compared to sunny days, and its performance varies between yards with different grass types. What is the primary testing challenge illustrated by this observation?

   • AThe mower's autonomy level is too low to handle different environmental conditions.
   • BThe mower's non-deterministic behavior makes it impossible to reproduce the same test results twice.
   • CThe mower's GPS navigation is inaccurate, causing it to behave differently in different yards.
   • DThe mower's performance is highly dependent on environmental conditions, making it difficult to ensure consistent and reliable operation across all real-world scenarios.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   What is the role of runtime monitoring in the verification and validation of autonomous systems?

   • ATo replace all other forms of testing
   • BTo continuously check the system's behavior during operation against safety properties
   • CTo ensure the system is used by authorized users only
   • DTo monitor the system's performance for optimization purposes
   Reveal Solution Discussion Select an answer first