agoeckner / autobot-racing Goto Github PK

Tasks

• Integrate control and guidance algorithms with the rest of the Autobot Racing system.
• Pass output of control systems to the vehicle transmitters.
• Test the complete integrated system with one vehicle.
• Test the complete integrated system with multiple vehicles.

Acceptance Criteria

• The full Autobot Racing system and the ControlSim simulator should use the same control/guidance algorithm codebase.
• Given a single vehicle test, the vehicle should successfully navigate the track.
• Given a multiple vehicle test, the vehicles should all successfully navigate the track, barring any collisions between vehicles due to a lack of overtaking capabilities in the guidance system.
• Given the complete integrated system, delay from computer vision vehicle identification to control system output must be negligible.

Assigned To: Anthony Goeckner
Workload: 10 hours

Tasks:

• Use OpenCV “contour” feature to determine the layout of the track.
• From the edges of the contour, generate a simpler polygon with a minimal number of vertices.
• Determine the navigable areas visible to the camera by finding both the inner and the outer boundaries of the track.
• Test the system in a variety of lighting conditions and track configurations.

Acceptance Criteria:

• Given a track with an inner and an outer boundary, the computer vision system will determine the navigable area.
• Given any reasonable indoor lighting environment (incandescent or fluorescent), the track will be accurately identified.
• Computer vision will only be used to identify the track during system initialization, at which time the layout will be stored for future use.

Assigned To: Zach Perry
Workload: 25 hours

Tasks:

• Test integration between each individual pair of components.

  1. CV -> Kalman Filter
  2. Kalman Filter -> Navigation, Guidance, Control
  3. Navigation, Guidance, Control -> Network
  4. Network -> Pi
  5. Pi -> Transmitter

• Perform overall integration testing of the entire system.

• This involves racing vehicles around the track in normal operating conditions.

Acceptance Criteria:

• All individual component integrations are tested successfully.
• Overall integration testing is performed successfully.
• The overall integration test is only considered a success if all component integration tests are successful.
• The overall integration test must be successful.

Assigned To: Anthony Goeckner, Ben Huemann, Zach Perry, Harold Smith
Workload: 20 hours (5 per person)

Tasks:

• Create message queue for main thread
• Create message queue for OpenCV thread
• Test that complete messages are received in a timely fashion.

Acceptance Criteria

• Given a well formed message when sent then the message should be received at the destination
• Given a well formed message when sent then the message should be received in a desired time
• Given a high volume of data traffic through the message-passing system, all messages will be delivered reliably and in a timely fashion

Assigned to: Harold Smith
Workload: 10 hours

Tasks:

• Create messages data type that is robust in format to allow for easy modification
• Create a way to allow PC to communicate to all Raspberry Pis
• Create a way to allow PC to communicate to individual Raspberry Pis
• Create a way to allow for Raspberry Pis to communicate with the PC
• Test message passing of each function for correctness
• Test message passing of each function for desired timing

Acceptance Criteria

• Given data on the PC when sent as a message to all Raspberry Pis then the same message will be received on every pi
• Given data on the PC when sent as a message to individual Raspberry Pis then the same message will be received only on that pi
• Given a reply on the Raspberry Pi when sent as a message to the PC then the same message will be received on the PC

Assigned to: Ben Huemann
Workload: 12 hours

Tasks

• Display simulated position and orientation of vehicle.
• Display desired course of vehicle.
• Display actual course of vehicle.
• Display a simulated course or navigable area.
• Test simulation program with a variety to control algorithms to determine correct operation.

Acceptance Criteria

• Given a simulation, the position and orientation of the vehicle are accurately displayed.
• Given a simulation, the desired course of the vehicle is accurately displayed.
• Given a simulation, the actual course of the vehicle is accurately displayed.
• Given a simulation, the navigable area of the track is accurately displayed.

Assigned To: Anthony Goeckner
Workload: 3 hours

Tasks:

• Create a virtual or physical line across the track
• Map the coordinates of the line in with the track boundaries
• Use OpenCV to outline the start line on the camera feed
• Test the creation of a start/finish line.
• Test recognition and mapping of start line

Acceptance Criteria:

• Given a start line, OpenCV will be able to detect and map the line
• Given a start line, the Camera feed will display, if virtual, the line on the camera feed
• Given a start line the a track and a start line, the two are differentiated by the computer vision system

Assigned To: Harold Smith
Workload: 10 hours

Tasks

• Map a coordinate system onto the image given by the camera.
• Test recognizing individual vehicles.
• Determine the orientation of each vehicle.
• Test recognizing the outline of the course.

Acceptance Criteria

• Given multiple vehicles in view of the camera, each can be uniquely identified.
• Given a moving vehicle, it can be identified as accurately as a vehicle sitting still.
• Given the computer vision system and a standard web-camera, a high image sampling rate can be achieved.
• Given an image containing vehicles, each vehicle can be precisely located on the internal coordinate system.

Assigned To: Anthony Goeckner
Workload: 20 hours

Tasks:

• Determine GPIO pins needed and initialize them
• Retrieve messages from communication socket
• Send to message data to control function
• Create three separate controlling functions
• Speed(float)
• Set appropriate pin to digital potentiometer to control Speed of RC car
• Turn(char)
• Set appropriate pin to turn the car left, center, or right
• Test speed function
• Test turn function

Acceptance Criteria

• Given message is received, then data is sent to the appropriate function
• Given the speed function is called, the RC car will change its speed according to data passed to it
• Given the turn function is called, the vehicle will turn its wheels

Assigned to: Ben Huemann & Zach Perry
Workload: 5 hours each, 10 total

Tasks:

• Initialize GPIO pins used to send power to controller
• Wire up RC controller to GPIO pins
• Power up controller in RCDriver initialization subroutine
• Power down controller in RCDriver deinitialization subroutine
• Test GPIO sends power to controller
• Test controller is initialized after digital potentiometer

Acceptance Criteria:

• Given GPIO is set to HIGH, the controller should power on
• Given GPIO is set to LOW, the controller should power off
• Given RCDriver is initialized, digital potentiometer should be set before controller is powered on

Assigned To: Ben Huemann
Workload: 15 hours

Tasks

• Use the vehicle target to determine the vehicle’s heading.
• Pass vehicle headings through a Kalman filter. This filter will first be implemented in another user story.
• Store a set number of previous vehicle headings, for use with later control systems.
• Test the system using a variety of vehicles at different locations on the track and pointed at different headings.

Acceptance Criteria

• Given a single vehicle in view of the camera, its position is accurately recorded.
• Given multiple vehicles in view of the camera, the position of each is accurately recorded.
• Given a moving vehicle, its position is updated with a high degree of accuracy.
• Given the computer vision system and a standard web-camera, a high image sampling rate can be achieved, allowing for a high-fidelity recording of the vehicle position over time.

Assigned To: Anthony Goeckner
Workload: 5 hours

Tasks

• Pick the library to use for developing the UI
• Draw out rough outline of desired UI and necessary components
• Begin coding the UI
• Test the functionality of each component in the UI

Acceptance Criteria

• A rough sketch of the main UI was drawn including all of the necessary components
• Half, or more, of the desired components were implemented
• Given a UI component when a user uses the component then the component will give the desired functionality

Assigned to: Harold Smith
Workload: 20+ hours

Tasks:

• Use a continuous instead of a discrete throttle setting.
• Change the speed parameter from an int to a float in the networking code.
• Test to make sure that no functionality or control has been lost.

Acceptance Criteria:

• Given an equivalent throttle level, cars must respond the same with the new system as they did with the previous system.
• All function calls must operate the same as before, error free.
• Given a value between -1.0 and 1.0, the system must set the throttle to that percentage of its maximum forward and backward (negative) speeds.

Assigned To: Zach Perry
Workload: 5 hours

Cars are not visible when they cross over the tape, due to lack of contrast.

Possible workarounds:

• Avoid tape by increasing wall distance.

Tasks:

• Determine a Kalman filter library to use.
• Determine and populate all initialization vectors of Kalman filter object
• Integrate filter input with position and heading data.
• Integrate filter output with existing navigation, guidance, and control systems.
• Test integration of the completed system.

Acceptance Criteria:

• Given noisy input position/heading data, the Kalman filter outputs a steady, reliable estimation of the true position.
• Given accurate input position/heading data, the Kalman filter output should not vary significantly.
• Given real time input position/heading data, the Kalman filter should provide a continuous stream of normalized position data

Assigned To: Anthony Goeckner, Ben Huemann
Workload: 20 hours (10 per person)

Tasks:

• Source track material
• Create track layout
• Mount camera to tripod
• Place Tripod in track

Acceptance Criteria:

• Given a track, the camera must be able to see the track without the tripod interfering with the track.
• Given a track, the material must be laid in a way that allows the cars to handle well
• Multiple track layouts must be provided to prove modularity and show that the tracks of the cars are not hard-coded or pre-determined

Assigned to: Zach Perry
Workload: 5+ hours

Tasks:

• Initialize each new car’s statistical values
• Track each individual car and detect car passing through start line
• Update and store new statistical values for the individual car
• Send updated values to the UI
• Test lap counting capabilities with multiple cars.
• Test lap counting when the car crosses different points of the start/finish line.

Acceptance Criteria:

• Given a new car, the car will have statistical values initialized to zero or None
• Given a car passing the start line, the counter for laps and lap times will be updated for the individual car
• Given a car passing the start line, the updated values for the individual car will be sent to the UI
• Given a set of updated values, the UI will update the values for the individual car in the “Leaderboard” frame

Assigned To: Harold Smith
Workload: 5 hours

Tasks:

• Perform testing in different lighting environments to determine computer vision failure modes.
• Make adjustments to CV tuning parameters to compensate for different lighting conditions.
• Research methods to automatically adjust CV parameter tuning based on environment.
• Test adjustments to ensure that the computer vision system works reliably in environments commonly found in the Lawson Computer Science Building.

Acceptance Criteria:

• Given multiple reasonable lighting environments (in LWSN), the computer vision system is robust enough to function in any of them.
• Given a proper lighting environment, the computer vision system is able to capture and track moving vehicles in all locations on the track.
• Given a proper lighting environment, the computer vision system tracks moving vehicles with minimal “drops”, or brief periods of time when vehicles cannot be detected.

Assigned To: Anthony Goeckner
Workload: 20 hours

Tasks

• Cause the computer vision system to poll the camera at a set frequency and determine the location of each vehicle.
• Pass vehicle locations through a Kalman filter. This filter will first be implemented in another user story.
• Implement a function to determine if a vehicle is within the boundaries of the track.
• Store a set number of previous vehicle positions, for use with later control systems.
• Determine a system response when vehicle cannot be located.
• Test the system using a variety of vehicles at different locations on the track.
• Test the system response when a vehicle cannot be found.

Acceptance Criteria

• Given a single vehicle in view of the camera, its position is accurately recorded.
• Given multiple vehicles in view of the camera, the position of each is accurately recorded.
• Given a moving vehicle, its position is updated with a high degree of accuracy.
• Given the computer vision system and a standard web-camera, a high image sampling rate can be achieved, allowing for a high-fidelity recording of the vehicle position over time.

Assigned To: Anthony Goeckner
Workload: 5 hours

Tasks

• Create a new version of the wall-following guidance system that allows for variable distance from the wall.
• Determine distance from one point on track to another. (Not the straight-line or Euclidean distance)
• Increase or decrease the vehicle’s distance from the wall when it is within a set track distance of the next car in front of it and its speed is faster than the car in front.
• Test the system using the ControlSim application developed in Sprint 1.
• Test the system using the real vehicles and track, if possible.

Acceptance Criteria

• Given an overtake situation where vehicle A is gaining on vehicle B, the distance from the wall of vehicle A is appropriately adjusted for passing.
• Given an overtake situation when the slower vehicle is on the inside of the track, the overtaking vehicle’s distance from the wall will be increased.
• Given an overtake situation when the slower vehicle is on the outside of the track, the overtaking vehicle’s distance from the wall will be decreased.

Assigned To: Anthony Goeckner
Workload: 20 hours

Tasks

• Implement API for control and guidance algorithms
• Implement Camera feed from OpenCV
• Implement Car Statistics Labels
• Implement Stop button pop up window
• Test the functionality of API
• Test the functionality of Camera feed
• Test the functionality of Car Statistics pop up window

Acceptance Criteria

• Given a Camera Feed the UI will display the feed live
• Given control and guidance algorithms, the user will be able to select the desired algorithm when adding or editing a car
• Given a set of cars driving on the track, the statistics label will display performance information for each car
• Given the stop button click event, a pop up window will appear notifying the user to put the cars back in start position

Assigned to: Harold Smith
Workload: 20 hours

Tasks:

• Write functions to connect the currently existing buttons.
• Write a start function to start a new race.
• Write a stop function to permanently stop a race.
• Write a pause function to temporarily stop a race.
• Write a resume function to resume a paused race.
• Test the start/stop/pause/resume functionality and buttons under various race conditions.

Acceptance Criteria:

• Clicking the “Start Race” button must start the race, causing cars to begin to move.
• Clicking the “Pause Race” button should temporarily stop a race. This stops the cars but the cars should resume from their current positions once the resume button is clicked.
• Clicking the “Resume Race” button should restart a previously paused race. Cars should continue from their paused positions, and the race will continue as normal.
• Clicking the “Stop Race” button must immediately halt the race, stopping the cars. This ends the race permanently, and cars must be reset before a new race can begin.

Assigned To: Zach Perry
Workload: 10 hours

Tasks:

• Change the guidance algorithms to increase speed of the cars when in a straight section.
• Change the guidance algorithms to return the cars to a normal speed and slow down at the end of a corner.
• Change the guidance algorithms to prevent on-track collisions and offs due to increased speed.
• Test new guidance algorithms.

Acceptance Criteria:

• Given a car enters a straight section of track, it should accelerate out of the previous corner before slowing down for the next corner.
• Given the cars are moving around the track, the cars should not crash or go off course.
• Given the cars are moving around the track, the cars should not accelerate into a car in front of them.

Assigned To: Zach Perry
Workload: 10 hours

Tasks

• Create a basic proportional controller to maintain correct heading.
• Create ability to induce error in simulated inputs.
• Apply simulated inputs to control system to determine response.
• Test output of control system for expected responses.

Acceptance Criteria

• Given some simulated error, control system is able to maintain a desirable course.
• Given a small amount of error, the control system will display a minimal amount of overshoot.
• Given a modeled control system, the output is easily comprehensible.

Assigned To: Anthony Goeckner
Workload: 7 hours

Tasks

• Research possible ways that the integral portion of a PID controller is implemented.
• Implement the Integral calculation and add to the existing proportional calculation.
• Test PID controller with computer vision input.

Acceptance Criteria

• Given some error, control system is able to maintain a desirable course.
• Given some error and a correction for that error, the control system will display a minimal amount of overshoot.
• Given input, normalized output should be better tuned than the just the stand-alone proportional mode.

Assigned to: Anthony Goeckner, Ben Huemann
Workload: 20 hours (10 per person)

Tasks:

• Create new vehicle tags with different identification colors
• Create enclosures for the control boards
• Test that vehicle tags are correctly identified by the existing computer vision system.
• Test that vehicle tags are durable and do not shift position or orientation while driving.

Acceptance Criteria:

• Tags must be durable.
• Tags must be visible to the CV system.
• They must reflect light evenly.
• They must each be individually colored for each car.
• Control enclosures must protect the boards while in storage and use.

Assigned To: Zach Perry
Workload: 10 hours

Tasks:

• Create main thread for UI and Framework
• Create thread for OpenCV
• Set locks on any shared values to maintain concurrency
• Test threads to ensure they run concurrently
• Test that threads run concurrently without errors

Acceptance Criteria

• Given a thread for UI and Framework, the thread functions as normal
• Given a thread for OpenCV, the thread functions as normal
• Given a function or variable accessed by multiple threads, a lock will be set on the function or variable to ensure concurrency between threads

Assigned to: Harold Smith
Workload: 10 hours

Tasks:

• Take apart each remote
• Measure & record voltages/resistance of the buttons, potentiometer, and power connections
• Desolder and remove buttons, potentiometer, and power wires from circuit board
• Solder new wires to striped buttons, potentiometer, power nodes so that the PI gains control of the transceiver
• Choose GPIO pins and connect the new wires to the pins
• Test conductivity between all new leads soldered to wires

Acceptance Criteria

• When the PI sends power to the power pins, the LED on the transceiver should light up signifying it is receiving power
• When the PI sends power to either the left/right buttons, the wheels on the RC car should turn accordingly
• When the varying current is sent to the potentiometer nodes, the car should accelerate, stop, or reverse according to current draw

Assigned to: Ben Huemann & Zach Perry
Workload: 5 hours each, 10 total

The OpenCV function cv2.imshow is extremely slow and drags the entire CV system down to single-digits FPS.

Tasks:

• Investigate hardware used in RC remote
• Test unmodified remote for current functionality
• Test feasibility of electrical connection
• Create plan of action to connect Pis to remotes
• Modify hardware to allow Pi to operate RC cars via software
• Create basic software driver to control remotes via software.

Acceptance Criteria

• Successfully attempt to connect Raspberry Pi with RC controller
• Given a Raspberry Pi, when connecting to an RC remote, the Raspberry Pi should have no hardware connection errors
• Given a software-controlled remote, the software should be capable of performing all operations originally available on the remote.

Assigned to: Zach Perry
Workload: 10+ hours

Tasks:

• Test computer vision against multiple cars (without track)
• Test computer vision against track (without cars)
• Test computer vision against cars and track

Acceptance Criteria:

• Given cars are tested independently, when tested for computer vision functionality the cars will be uniquely identified along with their orientation.
• Given track is tested independently, when testing for computer vision functionality the track should be identified along with its boundaries
• Given cars are overlaid on top of the track when testing for computer vision functionality, both should be identified independently

Assigned to: Ben Huemann
Workload: 10 hours

Tasks

• Create message passing functions for communication between classes
• Test message passing of each function for correctness
• Test message passing of each function for desired timing

Acceptance Criteria

• Given every class when trying to communicate between classes then each class should have the functionality to pass messages to one another
• Given a well formed message when sent then the message should be received at the destination with complete correction
• Given a well formed message when sent then the message should be received in a desired time
• Given a high volume of data traffic through the message-passing system, all messages will be delivered reliably and in a timely fashion.

Assigned To: Harold Smith
Workload: 7+ hours

Tasks:

• Create APIs for use by researchers to allow the use of researcher-designed control systems.
• Test the associated APIs.

Acceptance Criteria:

• An API is created which allows for multiple control systems to be used simultaneously, each on a different vehicle.
• The framework provides an interface to the camera to allow for computer vision.
• Given a certain specification for the structure of a control system, a researcher can create a control system that will interact successfully with the framework.

Assigned To: Harold Smith + Ben Huemann
Workload: 16 hours