Lesson 5 of 6 · Commands and autonomy
Give the robot a goal and boundaries
8 minute lesson · 3 checks along the way
Your goal for this lessonDistinguish a commanded sequence from bounded onboard decisions and explain the need for stop conditions.
In this lesson 4 sections · 3 checks
Using archive links & definitions
Words with a dotted underline open a short definition right here in the lesson.
Read & understand
A command can leave choices open
A fixed sequence requests specified actions in a planned order. A more autonomous instruction can state a goal and permit the robot to choose among actions using its sensors. Neither approach removes the need for tested software and operational constraints. The appropriate balance depends on uncertainty, available computation and the time needed for Earth to respond.
For a rover, people can choose a destination while onboard navigation evaluates local motion options. That division differs from asking a machine to invent its own scientific priorities without limits. Describe the actual decisions delegated rather than labeling every automated function as equally autonomous.
Check 1 of 3
10 pointsUse the reading above. You can try again without losing points.
Read & understand
A local loop can react before Earth
A sense-decide-act loop uses fresh local measurements to compare the state with the permitted behavior. It may stop, choose another candidate path or request later review when assumptions fail. A stop is sometimes the intended success of a protection rule rather than a failure to make progress.
Ground communication delay means an unexpected obstacle can develop before a new human instruction arrives. Onboard responses address that timing gap. Their usefulness depends on reliable sensing, well-defined authority and testing of both ordinary and unusual conditions.
Read & understand
Write a bounded fictional instruction
Consider: approach a selected observation area using evaluated terrain, keep within a defined resource budget and stop if location confidence or clearance is inadequate. This gives the robot an objective, constraints and failure behavior. “Reach the target at any cost” supplies none of the needed priorities.
The numerical thresholds in a real instruction must come from the vehicle’s tested capabilities. This lesson deliberately uses no flight limits. The reasoning exercise is to identify which variables need limits and what evidence the robot should save when it stops.
Autonomy has an authority boundary
Original teaching diagram. One fictional bounded navigation instruction. GOAL AND CHOICES. Reach the selected area. Choose evaluated paths. Use fresh local sensing. LIMITS AND EVIDENCE. Respect resource limits. Stop if confidence fails. Save the decision context. A protective stop can be successful execution of the instruction, not a failure of it.
Open full-size diagram in a new tab: Autonomy has an authority boundaryCheck 2 of 3
10 pointsUse the reading above. You can try again without losing points.
Read & understand
Supervision continues through returned evidence
A team reviews the route, sensor observations, rejected options and final state. That record helps determine whether an unexpected stop reflected an actual hazard, a conservative estimate or a sensing problem. A final photograph alone may not explain the decision.
Updating software or changing a limit can alter behavior outside the case that motivated the change. check previously working situations as well as the new case. Bounded is a continuing engineering and operations relationship, not a one-time transfer of responsibility to a machine.
Check 3 of 3
10 pointsUse the reading above. You can try again without losing points.
Put it together
Your lesson progress
Each check connects to a part of the story. Complete all 3 to earn your 25-point lesson bonus.
Explore further · Optional
Go to the source
Explore the evidence behind this lesson. These optional readings and videos open in a new tab.