RL Robotics LabSearch ↗
Lesson / authored

Week 24 · Experiment: Answer questions by tracing evidence

Session 3 of 4 · Demonstrate and defend · Phase 6

Plan about 15 minutes for explanation, 30 minutes for practical work and 10–15 minutes for documentation. A longer build may continue into the next session: stop safely, commit the current state and record the next check. Desktop simulations count as software evidence; label them clearly and record physical validation separately.

Engineering challenge

Can someone else understand, reproduce and challenge your engineering conclusions? This session focuses on answer questions by tracing evidence.

Before you start

The previous week’s recorded baseline and Week 23, Improve. For later sessions this week, retain the preceding session’s files and predictions.

Equipment: Desktop Python, editor, paper and ruler; for physical work, the configured 3pi+ 2040, clear floor mat and hardware checklist. Week 3 additionally uses the separate low-voltage LED circuit described in its procedure.

For any motion, verify the stop button, short time limit and clear floor area first. Keep the wheels raised for a new device program until its commands and stop behaviour are checked. A hazard or uncertain input is a reason to stop and document, not to force the trial to finish.

Theory and mathematics

Answer questions by tracing evidence

An engineering defence connects a claim to a measurement and its uncertainty. If asked why PID helped, show the baseline comparison rather than saying the code is more advanced. If asked whether a new obstacle is safe, describe the actual sensing limit and the stop policy. It is acceptable to say “not tested” and propose a concrete experiment. Separate statistical variation, calibration error and model limitations. A conclusion should remain within the conditions you evaluated, even if an ambitious extension is technically imaginable.

Worked example — illustrative values

A defensible claim is: “Version v1.0 arrived in 8 of 10 held-out trials on our surveyed mat; arrived endpoint error averaged 6 cm, with two hazard stops.” It names version, denominator, conditions and a conditional metric. “The robot always navigates accurately” is unsupported by that evidence. These figures are fictional examples, not course results.

Write the calculation in your notebook before running code. State which values you measured, which you assumed and which the program calculates. A correct numerical calculation cannot rescue an incorrect physical assumption.

Run and explain the model

The following is desktop Python, not a ready-to-run motor program. Download this week’s example, save it in your student repository and run python3 code/w24.py from the repository root. The same small model is reused across the week so you can learn it, build with it, test it and revise it.

# Desktop Python teaching example. Numerical inputs are illustrative.
def claim_summary(version, attempts, arrivals, conditions):
    if attempts <= 0 or not 0 <= arrivals <= attempts:
        raise ValueError("invalid trial counts")
    return (f"{version}: {arrivals}/{attempts} arrivals under {conditions}. "
            "See raw trials for errors, stops and limitations.")

print(claim_summary("example-v1", 10, 8, "fictional surveyed-mat tests"))

Use your own recorded data or named test fixtures instead of the illustrative inputs. Save expected and actual values side by side. If an exception appears, read its final line, identify the input or assumption that caused it and make the smallest explained correction. Do not delete validation merely to obtain output.

Understanding the model and its limits

The summary function checks that counts are logically possible before forming a claim. It deliberately leaves detailed errors and stop reasons in the linked raw record, because a one-line summary cannot establish every result. Replace its fictional version and conditions with the exact archived build and evaluated envelope. Then audit the prose: can each number be reconstructed, and are failed attempts represented? A peer should locate the dataset and repeat a calculation without asking you for an undocumented file. An honest “not tested” is stronger than a confident statement unsupported by evidence.

Practical instructions

  1. Ask a peer/adult to repeat setup from the README and replay one dataset.
  2. Present the five-minute defence and answer three evidence-based questions.
  3. Compare their measured endpoint with yours and explain any difference.
  4. Record questions you could not answer and propose exact follow-up tests.

Experiment

Have another person reproduce one result and challenge three claims; retain their questions and your evidence-backed answers.

Before testing, record your prediction, changed factor, measured response, fixed conditions and stopping rule. Save every attempted run, including failures, with a condition and source version. If hardware is unavailable, use an explicitly labelled synthetic/replay dataset and list the physical question it cannot answer. Do not invent completed trials.

Deliverable

An independent replay/check and three reasoned defence answers, with unknowns stated honestly.

Save notebook/w24-s3.md, the relevant code revision, raw CSV or test-case records, and one labelled diagram/plot/table. Link the files relatively from your notebook. Use the entry template and report guide.

Completion criteria

A documented failed prediction can meet the learning criteria. A missing physical trial must remain marked untested; software success alone does not validate the robot.

Reading and video

Reflection and next step

Which assumption most affected your result? Point to one observation that supports your explanation and one alternative explanation the evidence has not ruled out. Write a specific next test with a changed factor and measurable outcome, then proceed through the week’s Learn → Build → Experiment → Improve cycle.

← Previous sessionCourse roadmapNext session →