Description
Objective
This lab will teach you about probabilistic robot
localization. You will use: (1) Global Reference
Frame and Grid Cell Numbering and (2) World
Representation.
Requirements
Programming: Python
Robot: Webots 2023b (FAIRIS)
Sensors Utilized
To successfully complete this assignment, you will need to utilize several sensors on the robot. Each
sensor provides unique data essential for robust navigation and environmental awareness. Below is
an overview of the sensors, their specific functionalities, and references to previous assignments for
guidance on usage:
β’ Encoders: The encoders track the rotation of each wheel, enabling the calculation of distance
traveled and rotational changes. By using encoder readings, you can estimate the robotβs displacement and changes in orientation over time, aiding in precise position tracking. Refer to Lab
1 for an introduction to encoder usage.
β’ Compass: The compass sensor returns the robotβs orientation relative to the world frame, typically measured in degrees from the east direction (the x-axis). This sensor is essential for
maintaining and adjusting the robotβs heading during navigation tasks. Refer to Lab 1 for initial
setup and utilization of the compass.
β’ Lidar: The Lidar sensor captures a 360-degree range image of the environment by measuring
distances to obstacles and landmarks at various angles. This information allows the robot to map
surroundings, avoid obstacles, and make path-planning decisions based on spatial awareness.
Refer to Lab 2 to review Lidar integration and data handling.
β’ Camera with Object Recognition: The onboard camera, integrated with object recognition capabilities, allows the robot to detect and differentiate various objects in its environment. Specific
landmarks, such as colored cylinders, can be identified, providing visual cues for navigation and
goal localization. Refer to Lab 3 for implementation details on using the camera and object
recognition.
Each of these sensors provides critical information that will need to be integrated into your navigation
algorithm to ensure accuracy and efficiency. Be sure to review the respective labs to understand how to
utilize these sensors effectively as you design your solution.
Global Reference Frame and Grid Cell Numbering
The reference frame is set up as follows:
β’ The positive y-axis points North, while the negative y-axis points South.
β’ The positive x-axis points East, and the negative x-axis points West.
β’ The origin, (0,0), is located at the center of the arena.
The arena is divided into a 5×5 grid of 25 cells, numbered from 1 to 25. Each grid cell is a 1 x 1 meter
square, creating a structured layout for navigation and localization.
In each corner of the arena, there is a colored cylinder that serves as a landmark. These cylinders have
a fixed size:
Global Reference Frame and Grid Cell Numbering (cont.)
β’ Radius, r = 0.25 meters
β’ Height, h = 1.5 meters
β’ Each cylinder is centered at the corner of a grid cell.
The four cylinders are uniquely colored (yellow, red, blue, and green) to assist in visual identification
and orientation within the arena.
Figure 1. Global reference frame and grid layout for Task 1 (maze7.xml). The arena is
divided into a 5×5 grid with colored cylinders located in each corner. All units are in
meters.
Refer to this layout as you plan and execute navigation strategies within the defined grid.
Wall Configuration Data Structure
For this assignment, you will need to implement a data structure to represent the wall configuration
of each grid cell in the arena. This structure will allow your robot to recognize and navigate around
obstacles effectively.
As a conceptual guide, consider representing the wall configurations using a 25×4 matrix format, as
illustrated in Figure 2. In this example:
β’ The matrix corresponds to the 25 grid cells in the arena.
β’ Each cell in the matrix contains four entries representing the presence or absence of walls in the
West-North-East-South (WNES) orientation.
β’ A βWβ (wall) entry indicates the presence of a wall, while an βOβ (open) entry denotes the
absence of one.
While you are encouraged to design your own data structure, this example can serve as a reference to
help you get started.
Wall Configuration Data Structure (cont.)
Figure 2. Example of a wall configuration representation. The left side illustrates the
physical world, while the right side shows the corresponding 25×4 matrix representation
using WNES orientation.
Think carefully about how your data structure will allow you to efficiently check for walls in each
direction and update it as the robot moves through the arena.
Background: Localization with Landmarks and Normal Probabilities
In this section, we will introduce the concept of using landmarks for robot localization within a grid.
The robot uses distance measurements from known landmarks to estimate its location within a 5×5 grid
by comparing these measurements with precomputed distances from each grid cell to the landmarks.
For each grid cell CN, we compute the likelihood of the robot being in that cell based on the measured
distances to landmarks. This probability is calculated by comparing:
β’ DRβLi
: The distance from the robotβs current position to each landmark Li
.
β’ DCNβLi
: The precomputed distance from the center of each cell CN to each landmark Li
.
Using a normal (Gaussian) probability distribution function, we can calculate the likelihood of the robotβs position given these distances.
Calculating Normal Probability in Python
To calculate the probability density for a given distance measurement, you can implement the normal
distribution probability function using the following formula:
f(s) =
1
β
2ΟΟ2
exp
β
(s β Β΅)
2
2Ο2
!
where: – s is the mean (precomputed distance from the grid cell to the landmark), – Β΅ is the observed
distance (measured distance to the landmark), – Ο is the standard deviation, representing expected measurement noise.
Write a function in Python to calculate this probability. Here is the pseudocode to guide you:
1 # Pseudocode for calculating the normal probability density
2 def calculate_normdist (s, mu , sigma):
Background: Localization with Landmarks and Normal Probabilities (cont.)
3 “””
4 Calculate the normal probability density for a given state.
5
6 Parameters:
7 s (float): Precomputed distance from the grid cell to the landmark.
8 mu (float): The observed distance (measured distance to the landmark).
9 sigma (float): The standard deviation (expected noise level).
10
11 Returns:
12 float: The probability density value.
13 “””
14 # Step 1: Calculate the coefficient (1 / sqrt(2 * pi * sigmaΛ2))
15 coefficient = 1 / (sqrt (2 * pi * sigma Λ2))
16
17 # Step 2: Calculate the exponent (-((s – mu)Λ2) / (2 * sigmaΛ2))
18 exponent = -((s – mu)Λ2) / (2 * sigma Λ2)
19
20 # Step 3: Calculate the probability by multiplying the coefficient with the
exponentiated result
21 probability = coefficient * exp(exponent)
22
23 return probability
In this pseudocode:
β’ s represents the precomputed distance from a specific grid cell center to the landmark.
β’ mu represents the observed distance from the robot to the landmark.
β’ sigma represents the standard deviation, which models the expected measurement noise or uncertainty.
Example Calculation
To illustrate how to use this function, letβs walk through an example. Suppose the robot measures a
distance of Β΅ = 2.915 meters to a particular landmark, and the precomputed distance from a given cell
center to the same landmark is s = 2.828 meters. Letβs assume a standard deviation of Ο = 1.
Using the pseudocode function calculate normdist, we would calculate the probability as follows:
1 # Example values
2 mu = 2.915 # Measured distance to the landmark
3 s = 2.828 # Precomputed distance from cell center to landmark
4 sigma = 1 # Standard deviation
5
6 # Calculate the probability
7 probability = calculate_normdist (s, mu , sigma)
8 print(“Probability:”, probability )
In this example:
β’ The smaller the difference between s and mu, the higher the probability, indicating that the robot
is likely close to the grid cell associated with mu.
β’ This probability represents the likelihood of the robot being at the cell given the measured distance to the landmark.
You will use this function to compute the probability for each cell in the grid and determine the cell
with the highest probability as the most likely position of the robot.
Alternatively, you can use the Python Scipy libraryβs norm function to achieve this.
Background: Localization with Landmarks and Normal Probabilities (cont.)
1 # Import package
2 from scipy.stats import norm
3
4 # Parameters
5 mu = 2.915 # the measured distance to the landmark
6 s = 2.828 # the precomputed distance from a grid cell to the landmark
7 sigma = 1 # standard deviation
8
9 # Calculate the normal probability
10 probability = norm.pdf(s, mu , sigma)
11 print( probability )
Background: Localization Using Wall Configuration and Sensor Readings
In this section, we introduce the concept of using a sensor model to estimate the robotβs position within a
grid maze based on wall configurations. The sensor model, illustrated in Figure 3 (Figure 3), provides a
way to calculate the probability of the robot being in a given cell based on its surroundings. This model
relies on the robotβs ability to detect walls in four directions (north, south, east, and west) using LiDAR
or another distance-sensing device.
Sensor Model and Wall Configuration
The sensor model allows us to compare the observed wall configuration around the robot to known
wall configurations of each cell in the grid maze. Each cell has a specific layout of walls, which can be
represented in a data structure where each cellβs configuration is recorded. For example, a cell might
have walls on the north and east sides but open paths to the south and west. These configurations are
stored in advance, providing a reference for comparing with the robotβs observations.
Collecting Sensor Readings
To use the sensor model for localization, the robot must gather distance readings in four directions:
β’ North: The LiDAR measures the distance to any object directly in front of the robot.
β’ South: The LiDAR measures the distance to any object directly behind the robot.
β’ East: The LiDAR measures the distance to the right of the robot.
β’ West: The LiDAR measures the distance to the left of the robot.
These readings will indicate the presence or absence of walls in each direction (You will need to consider whether there is a wall based on a min and max distance for each cell), forming an observed
configuration of the surrounding area.
Calculating Probabilities Using the Sensor Model
Once the robot has collected LiDAR readings in all four directions, it compares this observed wall
configuration with the pre-stored configurations of each cell in the grid maze. The sensor model assigns a probability to each cell based on how closely the observed configuration matches the expected
configuration of that cell.
For each cell CN, the probability P(z = 0|s = CN) is calculated based on the match between:
β’ The actual wall configuration stored for CN, which specifies where walls are located around the
cell.
β’ The observed wall configuration derived from the robotβs LiDAR readings.
If the observed wall configuration matches the stored configuration for a given cell, that cell will have
a high probability, represented by P(z = 0|s = CN). If the configurations do not match, the probability
P(z = 0|s = CN) for that cell will be lower. This probability calculation is based on each directional
match.
Weighted Probability Calculation
Background: Localization Using Wall Configuration and Sensor Readings (cont.)
To determine the likelihood of being in a specific cell, calculate the weighted probability by multiplying the individual probabilities for each direction (north, south, east, and west) based on the match
between observed and expected wall configurations. The weighted probability for each cell is given by:
P(z = 0|s = CN) = P(znorth = 0|s = CN)Γ P(zsouth = 0|s = CN)Γ P(zeast = 0|s = CN)Γ P(zwest = 0|s = CN)
where each term P(zdirection = 0|s = CN) represents the probability that the observed reading in a specific
direction (e.g., north) matches the expected wall configuration in that direction for cell CN.
By calculating the combined probabilities P(z = 0|s = CN) for all cells in the maze, the robot can
determine the most likely cell based on the current wall configuration around it. This approach provides
a probabilistic localization method that leverages the robotβs immediate surroundings rather than relying
solely on distance measurements to landmarks.
Figure 3. Sensor model wall measurement probabilities for state probability calculations.
Task 1: Localization with Landmarks
In this task, you will estimate the robotβs position (pose) using landmarks and sensors, such as cameras,
IMU, or LiDAR. The positions of the landmarks are known in advance, and the robot will navigate
through an open-world environment (see Figure 4) starting from an unknown initial pose. You are not
allowed to use the GPS to determine the starting location. The goal is to estimate the robotβs pose
based on distance measurements to landmarks and navigate the entire grid, printing and comparing the
estimated positions as the robot moves.
β’ Pose Estimation Using Normal Distribution:
β For each grid cell CN, use a normal distribution function to estimate the robotβs pose by
comparing the robotβs measured distances to each landmark with the precomputed distances
from the cell centers to the landmarks.
β Measure the distances from the robotβs current position to each landmark using available
sensors (e.g., camera, LiDAR, IMU).
β Calculate the probability that the robot is in each cell by computing the weighted probability
as the product of normal probabilities based on the distance comparison:
WCN = NCNβL1 Γ NCNβL2 Γ Β· Β· Β· Γ NCNβLk
where NCNβLi
is the normal probability for cell CN based on landmark Li
.
β’ Identify the Current Cell Based on Highest Probability:
β Calculate the weighted probabilities for each cell each time the robot enters a new cell.
Task 1: Localization with Landmarks (cont.)
β Identify the cell with the highest probability WCbest and assume this cell represents the robotβs
current position.
β Print the estimated pose (coordinates of Cbest) each time the robot moves into a new cell.
This output should be displayed in the console and recorded in the video.
β’ Complete Coverage of All Grid Cells:
β Program the robot to navigate through the entire grid, ensuring that it visits each cell at least
once.
β Use a systematic approach, such as a lawnmower pattern, to ensure complete coverage
without unnecessary revisits.
β Keep track of the cells visited by assuming the highest-probability cell is the robotβs current
location.
β The task ends when the robot has successfully navigated all cells in the grid.
β’ Data Collection and Analysis:
β For each estimated position, obtain the actual (x, y) position from the GPS sensor for comparison (this can be used only for evaluation, not for navigation).
β Record both the estimated and actual positions for each grid cell entry and log this data.
β Use the collected data to create a graph comparing estimated positions to actual positions.
β Include a detailed analysis in the lab report, discussing any discrepancies between estimated
and actual positions and evaluating the accuracy of the localization method.
β’ Recording Requirements:
β Record a video showing the robot starting from at least two different initial positions and
navigating the grid to complete the task.
β Ensure that the console output of the calculated probabilities and the estimated pose for
each cell is visible in the video.
β The video should clearly demonstrate the robotβs movement, pose estimation calculations,
and grid coverage.
Figure 4. Open world layout for Task 1 with known landmarks for localization (maze
file maze7.xml).
Task 1: Localization with Landmarks (cont.)
Be sure to include a thorough comparison of the predicted pose versus the GPS readings in your
report, as this will be a key part of your evaluation.
Task 2: Localization with Walls and Landmarks
In this task, you will implement an algorithm that uses the sensor model to estimate the robotβs position
as it navigates through a 5×5 grid maze. The objective is to use LiDAR sensor readings to calculate
probabilities for each cell based on wall configurations, as described in the background section. The
robotβs starting location will not be known to the robot. This task will demonstrate how the sensor
model can aid in predicting the robotβs location, even though it may not lead to full localization due to
the symmetry of the maze.
The testing environment is shown in Figure 5, and the complete map configuration is known in advance to the robot. However, the robot will start from an unknown position in the maze, and students are
not permitted to use GPS or any other starting position attribute to determine the robotβs initial location.
β’ Navigating Through the Maze:
β Program the robot to navigate from cell to cell within the 5×5 grid maze. Each time the
robot enters a new cell, it should stop and take sensor readings in four directions: north,
south, east, and west, aligned with the cardinal directions.
β The robot should continue moving until it has entered at least 15 different cells. If the robot
encounters a dead end, it is allowed to backtrack to a previous cell and continue navigating.
β’ Map Data Structure and Sensor Model:
β Students must implement a data structure that stores the wall configurations for each cell in
the maze, aligning the wall measurements with the cardinal directions (north, south, east,
and west). This data structure will represent the known map of the environment.
β Each time the robot enters a new cell, it should use the sensor model to calculate the probability P(z = z
β²
|s = CN) for each of the 25 cells based on the observed wall configuration and
compare it to the stored wall configurations for each cell.
β Print the calculated probabilities for all 25 cells to the console each time the robot enters a
new cell. This output should be visible in the recorded video.
β’ Listing Cells with Highest Probability as Predictions:
β After calculating the probabilities for all cells, identify the cell(s) with the highest probability. If there are multiple cells with the highest probability (due to symmetry in the maze),
list all of them as the robotβs predicted position.
β Print the predicted cell(s) to the console as the robotβs best estimate of its current position.
This prediction should be visible in the video as well.
β’ Video Requirements:
β Record a video of the robot navigating the maze and performing the probability calculations.
The video should show the robot starting from multiple initial locations and moving through
at least 15 cells for each starting position.
β Ensure that both the probability calculations and the list of predicted cells are printed to
the console and clearly visible in the video. This will demonstrate the robotβs process of
estimating its position based on wall configurations.
β’ Note on Expected Outcomes:
β The purpose of this task is to illustrate how the sensor model can be used to predict the
robotβs location based on surrounding wall configurations. However, due to the symmetry
of the maze, the robot will not be able to fully localize itself using only the sensor model,
as multiple cells may have identical wall configurations.
Task 2: Localization with Walls and Landmarks (cont.)
Figure 5. Maze environment for Task 2, with known wall configurations for Bayesian
inference-based localization (maze file maze8.xml).
This iterative application of motion and measurement updates enables precise localization and navigation within the maze using Bayesian inference.
Extra Credit Opportunity (10 points)
For students who wish to attempt a more accurate localization, an additional 10 points of extra credit
will be awarded to the student who provides the correct grid localization for all grid cells in the shortest
time. To be considered, students must include a plot comparing predicted positions with GPS data and
record the fastest completion time in their report. Additionally, students need to explain exactly how
they implemented accurate predictions without ever using the GPS or starting location.
Code Evaluation Task 1 (45%)
β’ Probability Calculation Using Landmarks (10 points):
β Full Marks (10): The algorithm accurately calculates the probability for each of the 25 cells
based on measured distances to landmarks every time the robot enters a new cell, using the
normal distribution formula precisely.
β Acceptable (7-9): Probabilities are calculated for each cell, with occasional minor errors or
small deviations from the normal distribution formula.
β Poor (1-6): Probability calculations are implemented, but frequent errors or incorrect application of the normal distribution formula significantly impact accuracy.
β No Marks (0): Probability calculations are either missing or incorrectly implemented for
most cells, with no functional results.
β’ Printing and Displaying Most Likely Cell (5 points):
β Full Marks (5): The algorithm correctly identifies and prints the cell(s) with the highest
probability as the robotβs predicted position each time it enters a new cell. This output is
clearly visible in the video.
Task 1: Motion to Goal (cont.)
β Acceptable (3-4): The algorithm identifies and prints the most likely cell, with minor errors
or inconsistencies in visibility in the video.
β Poor (1-2): The most likely cell is occasionally identified, but with significant errors or
missing visibility in the video.
β No Marks (0): The most likely cell is not identified or displayed in the video.
β’ Using Highest Probability to Track Visited Cells (10 points):
β Full Marks (10): The algorithm accurately uses the highest probability to determine the
current cell and maintains an accurate record of visited cells.
β Acceptable (7-9): The algorithm tracks visited cells with occasional errors in determining
the highest probability cell or maintaining an accurate visited cells record.
β Poor (1-6): The algorithm tracks visited cells inconsistently, with frequent errors in determining the highest probability cell or maintaining a reliable record.
β No Marks (0): The algorithm does not track visited cells accurately, with the record incomplete or unreliable.
β’ Complete Grid Coverage (10 points):
β Full Marks (10): The robot successfully covers all cells in the 5×5 grid at least once,
demonstrating complete and systematic grid coverage.
β Acceptable (7-9): The robot covers most cells, with occasional missed cells or unnecessary
revisits.
β Poor (1-6): The robot fails to cover a substantial portion of the grid or exhibits frequent
unnecessary revisits, impacting systematic coverage.
β No Marks (0): The robot fails to cover most of the grid, with random or inconsistent
movement patterns.
β’ Plotting and Analysis (10 points):
β Full Marks (10): A clear and accurate plot is provided, comparing the robotβs predicted
positions (based on the highest probability cell) with its actual GPS-based positions. A
thorough analysis of discrepancies is included in the lab report.
β Acceptable (7-9): A plot is provided with a reasonably accurate comparison of predicted
and actual positions, with minor errors or a limited analysis in the report.
β Poor (1-6): The plot lacks clarity or has significant errors, with limited or unclear analysis
in the lab report.
β No Marks (0): The plot is missing or incorrect, and the lab report lacks any meaningful
analysis.
Deductions
β’ Print Statement Visibility (-20 points maximum): Up to 20 points may be deducted if required
print statements are not clearly visible in the submitted video. The severity of the deduction will
depend on the clarity and frequency of missing print statements.
β’ Not Showing Multiple Starting Locations (-10 points): A 10-point deduction will be applied
if the video for Task 1 does not show multiple starting locations.
β’ Robot Collisions with Wall (-5 points): A 5-point deduction will be applied each time the robot
collides with a wall, up to a maximum of 5 points.
Code Evaluation Task 2 (45%)
β’ Probability Calculation Using Sensor Model (15 points):
Code Evaluation Task 2 (cont.)
β Full Marks (15): The algorithm accurately calculates probabilities for each of the 25 cells
based on wall configurations every time the robot enters a new cell, correctly implementing
the sensor model formula.
β Acceptable (11-14): The algorithm calculates probabilities with occasional minor errors,
showing a generally correct but partially inconsistent use of the sensor model formula.
β Poor (1-10): Probability calculations are implemented, but frequent errors or major inaccuracies in the sensor model formula significantly impact the result.
β No Marks (0): Probability calculations are missing or incorrectly implemented for most
cells, providing no meaningful results.
β’ Printing and Displaying Predictions Based on Probabilities (10 points):
β Full Marks (10): The algorithm correctly identifies and prints the cell(s) with the highest
probability as the robotβs possible positions each time it enters a new cell, with clear and
consistent output visible in the video.
β Acceptable (7-9): The algorithm identifies and prints the likely cells, with minor errors or
occasional inconsistencies in the output visibility in the video.
β Poor (1-6): The probable cells are sometimes identified but with major errors or missing
output in the video.
β No Marks (0): The probable cells are not identified, or output is consistently missing from
the video.
β’ Robot Movement Through Non-Repeated Cells (15 points):
β Full Marks (15): The robot moves through at least 15 unique cells, revisiting cells only
when encountering dead ends, and adheres to a systematic movement pattern.
β Acceptable (11-14): The robot moves through 15 cells with minor, unnecessary revisits or
slight deviations from the systematic movement requirement.
β Poor (1-10): The robot does not move through at least 15 unique cells or frequently revisits
cells, failing to show a consistent or systematic approach.
β No Marks (0): The robot fails to reach 15 unique cells or shows random, repeated movement without any clear pattern.
Deductions
β’ Print Statement Visibility (-20 points maximum): Up to 20 points may be deducted if required
print statements are not clearly visible in the submitted video. The severity of the deduction will
depend on the clarity and frequency of missing print statements.
β’ Not Showing Multiple Starting Locations (-10 points): A 10-point deduction will be applied
if the video for Task 1 does not show multiple starting locations.
β’ Robot Collisions with Wall (-5 points): A 5-point deduction will be applied each time the robot
collides with a wall, up to a maximum of 5 points.
Report & Video Evaluation (10%)
A project submitted without a report, or without including the corresponding task video, will not be
graded and will receive a β0β for the complete lab.
Report Sections: The report should include the following (points will be deducted if anything is
missing):
β’ Mathematical Computations: Show how you calculated the speeds of the left and right servos
given the input parameters for each task.
β’ Conclusions: Analyze any issues encountered when running the tasks and discuss how these
could be improved. Discuss your results, including any navigation errors. Conclusions need to
Report & Video Evaluation (cont.)
reflect an insight into what was learned during the project. Statements like βeverything worked
as expectedβ or βI enjoyed the projectβ will not count as valid conclusions.
β’ Video: Upload the video to Canvas showing the robot executing the task (details below).
β’ Authorship Statement: Include and sign the following statement at the end of the report:
βI developed all the code and written report with the following exceptions:
[Student name: signature, date]
[Code sections (specify lines): references and additional comments]
If any external tools were used to generate code, you need to include the prompt and
output.β
Task 1: Landmark-Based Localization
β’ Show the robot moving from cell to cell within the 5×5 grid.
β’ Each time the robot enters a new cell, display the calculated probabilities for all 25 cells based
on distance measurements to landmarks. This information should be printed to the console and
visible in the video.
β’ Show the robotβs predicted position by printing the cell(s) with the highest probability. This
prediction should be visible in the video each time the robot moves to a new cell.
β’ Demonstrate complete grid coverage by having the robot traverse all cells in the 5×5 grid at least
once. The video should capture this full navigation path.
β’ Repeat the task from multiple starting locations.
Task 2: Sensor Model-Based Localization Using Wall Configurations
β’ Show the robot navigating through the 5×5 grid and entering at least 15 unique cells, only revisiting cells in cases of dead ends.
β’ Each time the robot enters a new cell, display the calculated probabilities for each of the 25
cells based on wall configurations observed in four directions (north, south, east, west). This
information should be printed to the console and clearly visible in the video.
β’ Show the robotβs possible position(s) by printing the cell(s) with the highest probability. This
prediction should be visible in the video each time the robot moves to a new cell.
β’ Repeat the task from multiple starting locations.
Note: Ensure that all probability calculations, cell predictions, and multiple starting locations are
clearly visible in the video for both tasks. This will demonstrate the functionality and robustness of your
localization algorithms across varied initial conditions.
Lab Submission
For Lab 4 there are two separate Canvas Assignments: Lab 4 Code Submission and Lab 4 Report Submission. Please follow the Submission policy outlined below. Note: Failure to adhere to the submission
policy will result in a reduction of 20 points.
Code Submission Policy
Students will need to upload a Python file (i.e. Webots controller) for each Task. The Python
file should follow this naming convention USF ID LabX TaskY.py, where X is the lab number
and Y is the task number. Example rockybull5 Lab1 Task1.py. Additionally, if a student has
created any functions that are not present in the controller but rather the MyRobot.py, the
student will also need to upload that python file as well using the following naming convention
USF ID MyRobot.py.
If custom functions are used in a studentβs submission and they are not in the controller or the
MyRobot Python files, an automatic Zero will be assigned for the code submission.
Each file should be uploaded to the same submission. DO NOT upload a zip file, Canvas allows
multiple uploads per submission. Failure to meet these submission policies will result in 20 points
off.
Report Submission Policy
Students will need to upload their videos to their USF OneDrive or USF Sharepoint and generate
a shareable link. Please ensure that you make it viewable to everyone or at the very least the TA
and the Professor (use their emails to add them to the access list).
Students will need to upload a PDF file of their lab report. Be sure that the lab report contains the
link to the videos and any Metrics requested by the Lab. These Metrics will be used to determine
which student receives the extra credit. Use the naming convention USF ID LabX Report.pdf.
Only PDFs will be accepted.
If the student used ChatGPT to generate code they will also need to provide the conversation used
to generate the code. This can either consist of screenshots of the conversation or a shareable link
that the system provides. This should be a separate document and not part of the report.
Each file should be uploaded to the same submission. DO NOT upload a zip file, Canvas allows
multiple uploads per submission. Failure to meet these submission policies will result in 20 points
off.
TA may ask you additional questions to gauge your understanding via Canvas Message or MS Teams.
Failure to reply may result in point deduction.