Elevator Motion

Open in new tab ↗

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 1 of 14
Elevator Motion: Acceleration, Velocity, and
Displacement Analysis
Measurement of Changes in Acceleration and Pressure
Scientific Inquiry with AI
Student Guide
Detailed Instructions & Course Material
Why This Lesson
This investigation uses an elevator as an experimental platform to explore motion and the
accelerometer and barometer in your phone as the measurement tools to interrogate that motion.
In this investigation you will learn how to use both the change in acceleration and air pressure to
determine how fast the elevator was moving and how far it travels. While the two sensors measure
very different physical properties, they both can provide very accurate measurements. As part of
the investigative process, you will also learn about using numerical integration, graphical analysis,
and how to design controlled experiments.
Throughout the investigation, you will direct an AI assistant to help you understand the phenomena
and analyze your data. The AI assistant will help you accumulate background understanding
through prelab discussions, build code to conduct computational analysis, walk you through
simulations to extend your understanding, help you make sense of the results, and assist you in
communicating your findings.
Learning Goals
Upon completion of this investigation, you should be able to:
Phenomena:
• Explain the relationship between acceleration, velocity, and displacement in both physical
and mathematical terms.
• Distinguish a constant sensor offset from random noise, describe the physical origin of each,
and explain how each will accumulate when performing numerical integration.
• Explain how two independent sensors measuring different physical properties can be used to
cross-validate velocity and displacement measurements, and describe possible explanations
for disagreement based on the sensor characteristics and the physical properties measured.
Experimental Practices:
• Direct an AI coding assistant to conduct numerical integration of time series data and use
graphical methods to visualize and validate the results. When appropriate, ask the AI to
explain the specific numerical methods used in the analysis, or to share the code it ran.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 2 of 14
• Design and carry out a control measurement that removes a drifting background quantity from
a difference measurement.
• Compare two independent measurements of the same quantity and provide plausible
explanations for the differences.
Habits of Mind:
• Test the output of any analysis — hand calculation, code, or AI-generated result — against an
independent check (known value, credible estimate, secondary independent measurement)
before reporting it.
• Identify when and how a control measurement can be incorporated into an experimental
design to improve the precision of a measurement.
What You Will Need
• A smartphone with the free Phyphox app installed, for its accelerometer and barometer.
• A computing device with access to an AI model.
• Access to an elevator serving 3–10 floors.
• A tape measure or meter stick, for Session B pressure measurements.
• A partner for Session B whose phone reports pressure at a rate close to yours. Android phones
often report about 25 samples/s and iPhones about 1 sample/s, and the control measurement
is easier to align in time when both records arrive at the same cadence. (If the rates differ, you
can use AI to resample both records to a common rate.)
The Core Idea
• Modern microelectromechanical systems (MEMS) accelerometers measure acceleration
precisely enough that velocity and displacement can be obtained from it by numerical
integration. This is the basis of inertial navigation, used in aircraft, drones, robotics, and game
controllers.
• In those systems, drift, which all sensors experience at some level, is the central engineering
problem; it is managed by combining the accelerometer with gyroscopes and with periodic
external references such as GPS. A constant offset b in the acceleration signal accumulates
as b·t in velocity and ½b·t² in displacement, so a small offset can dominate the displacement
result over a long record.
• An elevator ride supplies its own boundary conditions: the car is at rest before and after the
ride, and a round trip returns it to the starting floor. Those conditions supply known true values
against which the bias can be detected and the accuracy of the result evaluated.
• Barometric pressure provides a second, independent path to displacement through the
hydrostatic relation dP/dh = −ρg.
• Introducing a control measurement into an experimental design improves the precision of the
result; in the reference example of Table A1 the improvement is roughly 20×.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 3 of 14
Worked Example: What a Good Analysis Looks Like
A reference measurement for a round-trip elevator ride is shared in Table A1 (Appendix A). Refer to
that table to see what a clean result looks like at each stage of the analysis. Your own numbers will
differ. Treat the table as a model of process and magnitude.
Session A — Numerical Integration and Sensor Bias (~5 hours)
Session A establishes the central relationship of the lesson: acceleration integrates to velocity, and
velocity integrates to displacement. You record your own elevator ride, then direct an AI coding
assistant through that integration, diagnose the sensor offset that shifts the result, and correct it.
Working from acceleration → velocity → displacement reverses the displacement → velocity →
acceleration sequence most commonly taught in introductory physics. The direction used here is
the one modern motion control systems are built on, using MEMS accelerometers and gyroscopes.
Part 1 — Pre-Class Preparation 1.25–1.5 hours
The pre-class tutor prepares you with a working understanding of linear motion and the terms used
to describe it that you will use in class. You will also conduct the elevator ride measurement using
your smartphone, collecting the data that you bring to class for analysis.
A. Socratic AI Tutoring Session A (~25 minutes)
When conducting the prelab tutor, use the Scientific Inquiry with AI business account and
open the prelab tutor for Session A by clicking the “Open the custom GPT” button in the
prelab section on the knowledgewithai.com website. This will launch you automatically into a
tutoring session in ChatGPT. Confirm you are in “Chat” mode and not “Work”. These steps
will ensure the best tutoring experience and will automatically record the completion of the
tutoring session.
The AI tutor asks rather than explains: it opens with questions about your own experience riding an
elevator, moves to diagnostic questions on how acceleration, velocity, and displacement relate,
and introduces random noise and measurement bias so that the classroom analysis makes sense
when you get there. You do not need prior physics; the tutor will help you build an understanding
starting from your own personal background.
B. Prediction Exercise (~20 minutes)
Before making your acceleration measurement during your elevator ride, sketch the predicted
acceleration, velocity, and displacement curves for a ride that accelerates, cruises, decelerates,
and stops. Write this down in your notebook and bring it to class. Estimate the displacement you
expect for each floor the elevator travels (you can use AI to assess if your estimate is reasonable).

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 4 of 14
C. Record the Elevator Ride (~20–30 minutes)
Record one round trip spanning 3–10 floors: up
from your starting floor to the top floor of the trip,
then back down to where you began. Keep the
recording running while the car waits at the top
floor; do not stop and restart it, and do not touch
the phone. Secure the phone against a wall or
corner of the car with the y-axis aligned to the
direction of travel, so that it cannot move relative
to the car during the recording. Include at least 3 s
stationary before the ride and 3 s stationary after
it. Log the accelerometer (includes g), linear
acceleration, and barometric pressure channels
at the same time using Phyphox’s “Simple
Experiment” tool, at the maximum data collection
rate for each sensor (leave the rate field set to 0,
which Phyphox uses for the maximum rate). The figure below shows the Phyphox menu for adding
a “simple experiment”. For the upward trip in the elevator, the y-axis linear acceleration trace
should show a value near zero at rest, a positive pulse, a flat cruise back near zero, a negative
pulse, and a return to near zero. A mirror image of this will be recorded on the downward return.
Ideally, you do not want any additional stops on your ride, so it is best to do the experiment during a
time when the elevator is not busy. It is critical that you measure the acceleration for at least 3 s
before the elevator starts to move as well as after it stops. If you don’t get the measurement
started in time, repeat the experiment. Take a screenshot of your acceleration graphs, then export
the data using the three-dot menu (⋮) → Export Data → CSV (Comma, decimal point). Phyphox saves
a .zip file containing one comma-separated values (CSV) file per sensor; keep the whole .zip file.
Have the .zip file and the screenshot available for class.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 5 of 14
D. Review of In-Class Investigation (~10 minutes)
Skim the In-Class instructions so you arrive ready to work through the analysis yourself alongside
the class. Note which decisions you expect to make, such as: What values do I expect for velocity
and displacement? How will I determine whether an offset correction is needed? If it is, what steps
will I take? Bring the exported .zip file, your screenshot, your prediction sketches, and a laptop or
tablet.
Part 2 — In-Class Investigation 110 min
The class works through the analysis pipeline in parallel, each of you on your own measurement
data, with the instructor gating the stages.
A. Opening Exercise (~10 minutes)
• Using your whiteboard, sketch the shape of the acceleration curve for the ride you recorded,
then compare with a neighbor. Discuss what all the features represent in terms of your
motion.
• On your whiteboards, illustrate numerically how you can calculate the change in velocity that
occurs between two measurements of the acceleration by your sensor. You can assume that
the acceleration is changing slowly compared to the 100 Hz data collection rate.
Recall: 𝑎 = ∆"
∆#
• Sketch the shape of the velocity curve for the entire ride, then compare with a neighbor.
• On your whiteboards, illustrate numerically how you can calculate the change in displacement
that occurs between two sequential velocity measurements. You can assume that the velocity
is changing slowly compared to the 100 Hz data collection rate.
Recall: 𝑣 = ∆$
∆#
• Sketch the shape of the displacement curve for the same ride, then compare again.
B. Co-Investigation with AI (~70 minutes)
Working with an AI coding assistant, you build the integration pipeline in four stages to analyze your
linear acceleration data. This channel reports acceleration with the gravitational component
already removed by your phone’s sensor fusion (it uses multiple sensor inputs and software to
make the correction for gravity), so it can be integrated directly and the residual sensor offset is the
only constant error in the record. Reference values from a measurement made on a typical
elevator appear in Table A1 (Appendix A) and are available as magnitude checks at every stage.
Stage 1 — Load and inspection (~10 minutes)
Upload the .zip file (or the linear acceleration CSV file inside it) in your chatbot window and prompt
the AI to plot the y-axis linear acceleration against time. Compare the graphs to your screenshot to
verify the data has been properly transferred before moving to the next step of analysis.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 6 of 14
Stage 2 — Integrate once, without correction (~20 minutes)
Instruct the AI to integrate the linear acceleration data, with no correction, to obtain velocity, and to
integrate the velocity to obtain displacement. Read the final velocity from the graph. The car is at
rest at the end of the ride, so the true value is 0 m/s. If the velocity is not zero at the expected
times, there must be an error in the measurement or the analysis. Evaluate the round-trip
displacement using the same approach: the car returned to its starting floor, so the true value is
0 m. Record the values of the velocity and the displacement at the end of the elevator ride based
on this initial analysis. Share your values using your whiteboards.
Consider how a small bias in the acceleration measurement would impact the analysis. Take a
moment to estimate how a small error might add up over time (e.g., if you accelerate at 0.2 m/s2
you will be moving at 2 m/s after 10 s). What average bias error would be required to produce the
error you saw at the velocity of your analysis? Discuss in groups and conduct the analysis
manually.
Now consider the difference between bias and random noise. How would random noise on the
acceleration measurement centered around zero impact the analysis?
Stage 3 — Determine and correct for bias, then integrate: acceleration → velocity →
displacement (~20 minutes)
Compute the mean value of the acceleration over the stationary segment at the start of the ride.
The mean value is your offset estimate for the linear acceleration data.
Instruct the AI to subtract the calculated offset from the whole acceleration record to create a
corrected acceleration, then integrate the corrected acceleration to obtain velocity, and integrate
again to obtain displacement. Plot the corrected acceleration and the calculated velocity and
displacement on the same graph, using a secondary axis for displacement if needed to show scale.
Stage 4 — Validate (~20 minutes)
Confirm that the corrected velocity returns to 0 m/s when the elevator is known to be stationary
and that the corrected round-trip displacement approaches 0 m. Compare the peak velocity and
the one-way displacement against your own estimates, using any background information you can
identify. Then ask the AI to refine the offset estimate so that the displacement remains constant to
within 2 cm over the stationary segment at the end of the measurement, and to explain the general
approach it used to do the refinement. Create a final graph of the acceleration, velocity, and
displacement.
C. Peer Comparisons (~10 minutes)
Compare your peak velocity, total displacement, and offset magnitude against other students’
measurements, looking for classmates who used the same elevator. Offsets differ from phone to
phone because each accelerometer carries its own factory calibration residual, and because the
sensor fusion that removes gravity from this channel leaves a small residual that depends on how
the phone was oriented in the car. Share findings with the entire class.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 7 of 14
D. Synthesize the Physics (~10 minutes)
Whole-class discussion returns to the core relationship. Velocity is the running integral of
acceleration and displacement the running integral of velocity, so a systematic error in the
measured quantity is amplified by each integration: as t in velocity, and as t² in displacement.
Reflect: Where Does It Break Down? (~10 minutes)
Write a short reflection on what the uncorrected and corrected displacement values showed you
about the precision of raw sensor data, and on how small variations in a sensor bias might affect
the use of accelerometers in different applications.
Deliverables — Session A
• Retain your conversation so you can build the post-lab report from it.
• Submit your final graph of acceleration, velocity, and displacement from Stage 4 on the
knowledgewithai.com website at the end of class.
Part 3 — Post-Class Work ~1.75 hours
Post-class work lets you consolidate and summarize the in-class analysis. It also extends the
investigation to elevator rides with different operating parameters, which is where you build on your
understanding, your mathematical thinking, and your graphical analysis skills.
A. AI-Generated Report (~45 minutes)
Prompt the AI assistant for a short report describing the pipeline: how acceleration became
velocity, how velocity became displacement, and where the offset correction entered. The report
should include all the intermediate and final graphs that help tell the story of the analysis, and the
three simulation records from B below as an appendix. Check the accuracy of every number and
graph; the responsibility for validation is yours.
B. Extension Activity: Prediction Simulation (~60 minutes)
An AI was used to create a simulation for you to extend your understanding and build your problem-
solving skills. Download Elevator_Motion_Prediction_SimulationA.html from the
knowledgewithai.com website and open it in any web browser; no installation or account is
needed. Each simulated ride is a single one-way trip: 3 s at rest, ease in, build speed, hold a cruise,
ease out, then 3 s at rest. The controls are floors traveled, floor height, direction, cruise speed,
maximum acceleration, jerk — the rate at which the acceleration itself changes — and sensor
noise. Open the application, explore how it works and how changing the parameters changes the
motion graphs. If you reveal the graphs and then move the slider bars, you can see how each
variable changes the motion graphs.
The application is designed to allow you to save your sketch and then reveal the simulated graphs.
Each export is a single PDF carrying the ride settings, the locked prediction sketch, and the

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 8 of 14
revealed curves. The ride report at the end of the file is written only when the graphs have been
revealed, so reveal before exporting. Include three exported records in the appendix of the report
from A above. Export the PDF immediately after revealing, before you move any slider: moving a
slider after the reveal clears your sketch.
Record 1 — Prediction of User Defined Elevator Ride (~20 minutes)
Set the parameters for a ride, click “Sketch my prediction,” and sketch the predicted curves on the
locked axes of both tabs (Acceleration & Jerk; Velocity & Displacement). Then reveal and export.
The sketch tool locks at the moment of reveal, so the sketch in the exported PDF is the one you
committed to beforehand.
Record 2 — Prediction of Random Elevator Ride (~20 minutes)
Use “New Random Conditions” to create a new ride. Carefully sketch your predictions on both
tabs before revealing the graphs. Export your PDF immediately after revealing, before moving any
slider.
Record 3 — Match your own ride (~20 minutes)
Set the controls so the simulation reproduces the elevator ride you measured and analyzed in
class, then export the third record. The simulation models a one-way trip, so match the upward leg
of your ride and set Direction to Up. Adjust the settings to reproduce the graphs from your analysis:
• Floors traveled. Counted during the ride.
• Average floor height. The one-way displacement from your Session A analysis divided by the
number of floors.
• Cruise speed. The height of the flat stretch in the corrected velocity curve.
• Maximum acceleration. The height of the acceleration plateau, read off the corrected
acceleration trace.
• Jerk. The plateau height divided by the time the trace takes to climb from zero to it, which is
the slope of the acceleration during the ease-in.
• Sensor noise. Roughly the standard deviation of your acceleration during the stationary
segment before the ride (ask the AI to compute it).
Set the Accelerometer reading toggle to “Gravity removed” so the simulated trace sits on the same
footing as the linear acceleration channel you analyzed in Session A. Compare the simulated
curves against your measured ones and account for the differences.
Deliverables — Post-class Session A
Nothing is submitted after Session A. Keep your report, with the three simulation PDF records in its
appendix. After Session B you will extend this report and submit the combined version.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 9 of 14
Session B — Alternative Measurement Modality using Pressure
Sensor (~5 hours)
Session B starts by analyzing the “acceleration (with g)” time series data using the workflow you
developed in Session A. That serves as verification and reinforces the analysis method. You then
use your barometric sensor data to calculate the change in altitude and the maximum velocity of
the elevator. Converting a pressure change into an altitude change first requires determining the
relationship between the two variables at your location. In addition, you will develop an
experimental design that uses a second phone as a control to correct for small changes in the
background pressure that occur within your measurement period.
Part 1 — Pre-Class Preparation ~1 hour
The pre-class work gives you the basic physics and nomenclature of pressure, and leads you
through a discussion of how controls are used in experimental design.
A. Socratic AI Tutoring Session B (~25 minutes)
When conducting the prelab tutor, use the Scientific Inquiry with AI business account and
open the prelab tutor for Session B by clicking the “Open the custom GPT” button in the
prelab section on the knowledgewithai.com website. This will launch you automatically into a
tutoring session in ChatGPT. Confirm you are in “Chat” mode and not “Work”. These steps
will ensure the best tutoring experience and will automatically record the completion of the
tutoring session.
The tutor is focused on helping you learn four topics:
• Using control measurements in experimental design (~12 minutes), opening with everyday
examples and narrowing to a pressure measurement: a single-phone measurement at two
elevations drifts with the background pressure of the room, while a two-phone measurement
holds one phone at a fixed reference elevation as the second moves.
• Pressure as a physical quantity: what it measures, the units used to report it, and the
magnitude of the force the atmosphere exerts on a person.
• The hydrostatic relation dP/dh = −ρg and the conversion constant it supplies.
• The barometric channel as an independent check on an accelerometer displacement
estimate.
If you would like to review the key information later, ask the AI to prepare a one-page summary of
the session.
B. Estimation of Displacement using Pressure (~25 minutes)
Make a rough measurement of how pressure changes with elevation, using the Pressure tool in
Phyphox (under Raw Sensors): hold the phone still on the floor for about 10 s and record a typical
value of the pressure, then hold it still at a height of about 2 m for another 10 s and record a typical

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 10 of 14
value again. Use the difference to determine the relationship in Pa/m. Then, using your own
measured value, answer this question: given the displacement you measured in Session A from
acceleration, what pressure change would you expect to find in the data you recorded during your
elevator ride? Be prepared to show that calculation at the start of class — don’t worry about the
accuracy at this point, just the concept. You will make a far more accurate measurement of this
relationship in class.
C. Review of In-Class Investigation (~10 minutes)
Review the investigation you will conduct in class. Try to identify a partner whose phone has a data
collection rate similar to yours. If your team has two different data collection rates, you can still
work together but will need to get help from your AI assistant to create evenly spaced data for
analysis. Bring the .zip data file from your Session A elevator ride; its barometer record is analyzed
in class.
Part 2 — In-Class Investigation 110 min
You start by analyzing the acceleration (with g) versus time data using the workflow you established
in Session A. You then work in pairs to determine the hydrostatic relationship using an
experimental design with a control measurement. That relationship converts your pressure versus
time data into displacement versus time. You fit the slope of the displacement graph during the
constant-velocity cruise to determine velocity, and finally compare the displacement and velocity
obtained from the two different physical measurements.
A. Activation (~15 minutes)
• Your instructor sets conditions using the Elevator_Motion_Prediction_SimulationA.html, and
the class commits on whiteboards to how the acceleration, velocity, and displacement curves
will change before the reveal.
• Students share the results of their pre-class measurement of the hydrostatic relationship.
• Instructor leads an investigation of how data sampling rates can limit experimental
investigations.
B. Co-Investigation with AI (~75 minutes)
The investigation runs in four stages.
Stage 1 — Acceleration (with g) cross-check (~20 minutes)
Use the Session A workflow on the acceleration (with g) channel. This channel reports proper
acceleration, so a stationary phone with one axis vertical reads about 9.8 m/s² on that axis rather
than zero. Use the workflow you developed in Session A to correct for the acceleration offset,
determine velocity and displacement, and create a motion graph which includes all three
variables. Compare the maximum velocity and total displacement you obtain against your linear
acceleration results from Session A, and account for any difference.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 11 of 14
Stage 2 — Measure without a control (~15 minutes)
Working in pairs with a single phone, measure the pressure at two elevations, using the largest
height difference you can measure accurately to ±1 cm. Repeat the pair of readings 2 minutes
later. Share results with the class on whiteboards and discuss the origin of the variation.
Stage 3 — Add the control (~25 minutes)
Repeat the same two-elevation measurement using two phones logging pressure at the same time.
One phone stays at a fixed reference elevation for the whole measurement while the second is
moved to the target elevation; repeat this several times within the same experimental data set so
you have multiple measurements to work with. (Consider what happens if you do not start the
phones together.) Describe your experiment to an AI and determine the best way to calculate the
change in pressure per meter using your full data sets from the two phones. Once you decide on a
method, ask the AI to perform the analysis and provide a graphic illustrating the results of the
analysis. Be prepared to share your method with the class. Compare values across the class on
whiteboards (report in Pa/m).
Stage 4 — Apply the measured constant to the elevator ride (~15 minutes)
Use AI to create a pressure vs time graph for your barometer data collected during the elevator ride.
Convert the pressure-against-time data into altitude using the constant you measured in Stage 3.
Does your height reach a plateau at the end of the ride? How does it compare to the height you
determined from your acceleration data? Determine velocity from the slope of the altitude trace
during the constant-velocity cruise phase. Create graphs showing the fits and the slope values.
C. Critique and Reflect (~20 minutes)
• Use AI to create a comparison table showing the height and maximum velocity for all three
data sets. Write a short discussion comparing them and say why you might have more trust in
any one of them where they differ.
• Using your own numbers: how much did adding the control change the spread in your
measurement, and what physical quantity did it remove?
• In most science classes a control means holding a variable constant. Background pressure is
not something you can hold constant. Explain what your control did instead, and why that
works.
• Name a measurement where this approach would fail — where a second, simultaneous
reference would not remove the disturbance — and say what makes it fail.
Deliverables — Session B
• Retain your AI conversation for the construction of the post-lab report.
• Submit the comparison table for all three data sets on the knowledgewithai.com website
at the end of class.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 12 of 14
Part 3 — Post-Class Work ~2.25 hours
Post-class work audits the cross-validation, extends the investigation of correcting for offset bias
using a simulation, and connects the pressure-altitude relationship to a system deployed
commercially.
A. AI-Generated Report (~60 minutes)
Combine the results from Session B into the report you created for Session A by prompting the AI
assistant to merge them. Check all the information for correctness — nothing should be in the
report that you did not read and verify. Include key graphs, your calculation of the hydrostatic
relationship, and the final table comparing the results for all three methods. Include your
assessment of the three methods, what you trust, and what might have caused differences if they
exist. Write that final summary yourself rather than generating it with AI.
B. Extension Activity: Postlab Simulation (~20 minutes)
Download Elevator_Bias_and_Motion_SimulationB.html from the knowledgewithai.com website
and open it in any web browser. This simulation is different from the simulation tool you used after
Session A. Every generated ride carries a new random sensor offset that stays hidden until it is
revealed, so you locate it the same way you did with your own recording in Session A. Generate a
ride, enter your offset guess, and then reveal the true bias. Add two screenshots to your report as
an appendix: (1) the Correct the Bias tab showing your offset guess and the revealed true bias, and
(2) the Error Propagation tab showing how a leftover offset error grows in velocity and
displacement.
C. NextNav Technology and Business Evaluation (~30 minutes)
Research the Pinnacle barometric vertical-location service operated by NextNav, which
determines which floor of a building a 911 caller is on using the pressure-altitude relationship from
this lesson. The Federal Communications Commission requires wireless carriers to deliver vertical
location within ±3 m for 80% of qualifying calls, and Pinnacle meets that requirement through a
network of fixed reference stations that model local pressure conditions — the Stage 3 control
measurement operated at metropolitan scale. Consider: how might pressure-based altitude
measurement be useful in emergency response, and what does that suggest about the value of the
NextNav technology and business model? You may use AI to help assemble the analysis, and you
are encouraged to draw on academic literature. Be prepared to speak to your conclusions in the
next class session.
D. Retrieval Practice / Course Review (~30 minutes)
Spend approximately 30 minutes on integrated course review using the customized AI-enhanced
retrieval tool, which supports spaced repetition and interleaving across the full course. Items
drawn from this lesson review both sessions and cover integration, offset correction, control
measurements, and cross-validation.

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 13 of 14
Deliverables — Post-class Session B
The bold items should be submitted using the knowledgewithai.com website.
• Combined Session A and B report described above in Part 3A, with the Simulation B
screenshots from Part 3B in its appendix.
• Notes on NextNav technology to discuss in class.
• Retrieval Tool Practice

Scientific Inquiry with AI • Elevator Motion • Student Guide
Lawrence Livermore National Laboratory | Page 14 of 14
Appendix A — Reference Measurements
Table A1. Reference measurements that can serve as sanity checks on magnitude and process.
Every building, elevator, and phone placement produces different results, and your values will
differ.
Session Quantity Reference value
A Ride configuration 5-floor round trip
A Peak velocity ≈ 1.83 m/s upward; ≈ −1.83 m/s downward
A Uncorrected round-trip
displacement
≈ 17.86 m (true value 0 m)
A Offset refinement ≈ 5 × 10⁻⁴ m/s² change to the first-pass offset estimate from
the stationary segment
A One-way displacement after
refinement
≈ 17.91 m one way, not verified against an independent
measurement of the floor-to-floor distance; residual round-
trip displacement under 2 cm
B Uncontrolled pressure
measurement
Single phone, Δh = 1.716 m, read at two different times:
20.6 Pa and 21.4 Pa
B Controlled pressure measurement Same Δh = 1.716 m with a fixed control phone: a repeatable
19.0 Pa (≈ 11.1 Pa/m)
B Larger-scale control example Corrected precision improved by roughly 20× (0.389 hPa
uncorrected vs. 0.018 hPa corrected)
B Acceleration (with g) cross-check Consistent with the linear acceleration result to within a
percent
B Barometric velocity (self-
measured constant ≈ 11.1 Pa/m)
≈ 1.79 m/s upward; ≈ −1.77 m/s downward, consistent with
the accelerometer values