CS 7638 - AI for Robotics - Mini Project: PID Solved

CS 7638 - AI for Robotics - Mini Project: PID

Introduction
This projects involves creating a Proportional-Integral-Derivative (PID) controller that: * maintains a simulated turbo pump * controls a rocket’s thrust

Project Description
The goal of Rocket PID is to give you more practice implementing your own Proportional-Integral-Derivative Controller that you learned about from the PID lesson. The main objective of this project is to write your own PID controllers to solve certain problems that are more easily solved using such controllers. The project is split into three sections, Part A, Part B and Part C. It is suggested that you use a PD controller to solve Part A and a PID controller to solve parts B and C but any approach that can solve the student testing suite will be accepted. You may also use Twiddle to find an optimal solution, but it is not required. This project requires the SciPy library and can display a graph using turtle or optionally the Matplotlib library.

PART A:

A liquid rocket engine requires precise control of the pressure for the feed system of liquid oxidizer and fuel into the engine turbopumps. A failure of engine turbopumps to supply the pressure to meet feed demands of the engine could force an abort of the launch or a disaster after liftoff. The main purpose of Part A will be to solve the problem of maintaining the fuel mixture pressure supplied by the rocket’s turbopumps. In order for the rocket to leave the launchpad, the pumps must be pressurized up to their max required level as quickly as possible. Once there, the pressure must be maintained for the duration of the launch. Therefore, the goal will be to create a simple PD based controller that can make adjustments to the pump’s output in order to meet pressure demands.

Your pressure tank starts initially at a level of 10 units. A constant consumption rate of 5 units is applied during this entire test (i.e. consumption does not change over any operating conditions and the system is always losing 5.0 units per time step). This over two time steps, the pressure level will fall to zero and if the pressure continues to fall below zero, you will fail the test. To counteract this, you have control of a pressure feed that supplies or removes pressure from the system. Your pressure feed initially starts at zero but you can adjust the flow from the feed system in increments of +/- 1.0 pressure unit per time step. Thus, after three time steps of adjusting your pressure feed by +1.0 units, your pressure feed would be supplying +3.0 units of pressure to the system per time step. A final constraint on your feed system is that it is only capable of a maximum rate of change in system pressure of +/- 10.0 pressure units per time step. Overall, this dynamic between a constant consumption and your feed system results in your controller being capable of altering the system pressure in the range of -15.0 to +5.0 units of pressure per time step. For example, if the current system pressure was 95 and your feed system was supplying 10.0 units per time step, the overall system pressure change would be from 95 to 100 in the next time step. Thus, it is impossible to increase the pressure more than 5.0 units or to fall by more than 15.0 units in a single time step. However, if the turbopump current pressure goes below 0 or above 105, it will result in damage to the engines and failure of the test.

For Part A complete the “pressure_pd_solution” in RocketPIDStudent_submission.py. To test your pressure PD controller, see the “test_running_pumps” test case located in “RocketPID_tester.py.

PART B:

The second part of the assignment is a slightly more difficult adaption of the PID controller for controlling a rocket launch and reentry. It requires your controller be able to control the output of rocket engines on a simulated rocket ship such that it is able to successfully maintain two specific velocity profiles through different atmospheric flight regimes as parameters such as thrust, weight, and air drag change over the time of flight. You will be given an evaluation file that you can use to determine how well your PID solution and parameters are working.

The controller can only operate the throttle on the rocket’s engines in a range of [0,1] and the rocket engine only pushes the rocket upwards (i.e. only gravity pulls the rocket downward). When enough throttle is applied, the rocket will “liftoff” and continue to ascend until it runs out of fuel or “lands”. The flight plan is set in the Test Cases. Once the rocket has completed its flight plan, it will be required to descend through gravitational forces and NOT by using the rocket engines (which only act in the upward direction and can only be used to slow the rocket’s descent on reentry). The rocket should then “land” by approaching the ground below a “safe” landing velocity of 0.1 km/s +/- 0.01.

For Part B, complete “rocket_pid_solution” in RocketPIDStudent_submission.py that satisfies the given constraints. To test your PID controller, use the test case “test_rocket_flight” in the RocketPIDStudent_submission.py file to import your throttle controls and test it on a simulated rocket.

PART C:

The third part of the assignment is an extension of Part B. The rocket is a bi-propellant liquid rocket, which uses a liquid fuel as well as a liquid oxidizer. Both the fuel as well as the oxidizer will be mixed in a fixed ratio, and the excess fuel or oxidizer will be wasted. This part requires two controllers, one to control the output of fuel and another to control the output of oxidizer such that it is able to successfully maintain one specific velocity profile through different atmospheric flight regimes as parameters such as thrust, weight, and air drag change over the time of flight.

The controller can only operate the throttle for the fuel and the throttle for the oxidizer on the rocket’s engines, both in a range of [0,1] and the rocket engine only pushes the rocket upwards (i.e. only gravity pulls the rocket downward). When enough throttle for each propellant is applied, the rocket will “liftoff” and continue to ascend until it runs out of fuel or runs out of oxidizer or “lands”. The flight plan required for the rocket consists of maintaining an upward velocity of 0.25 km/s +/- 0.01 for 130 seconds for period identified in the velocity profile. Once the rocket has completed its flight plan, it will be required to descend through gravitational forces and NOT by using the rocket engines (which only act in the upward direction and can only be used to slow the rocket’s descent on reentry). The rocket should then “land” by approaching the ground below a “safe” landing velocity of 0.1 km/s +/- 0.01.

Note that it differs from Part B in two aspects. One is that here the controller needs to operate the throttle of the oxidizer in addition to the throttle of the fuel. Second is that the flight plan consists of maintaining only one upward velocity unlike Part B which required maintaining two upward velocities.

For Part C, complete “bipropellant_rocket_pid_solution” in RocketPIDStudent_submission.py that satisfies the given constraints. To test your PID controller, use the test case “test_bipropelleant_rocket_flight” in the RocketPIDStudent_submission.py file to import your throttle controls and test it on a simulated rocket.

Visualization Options
Various information about the pumps, rocket’s flight path, and other telemetry can be viewed by enabling graphing using the “–showgraph” argument. By default it will try to use matplotlib if installed or will use turtle graphics otherwise. To specify which version you want to use, “–showgraph turtle OR –showgraph matplotlib”.

Testing Your Code
We have provided a testing suite similar to the one we’ll be using for grading the project, which you can use to ensure your code is working correctly. You are guaranteed the points on the project you receive from the testing suite if you satisfy the tester AND the conditions laid out for passing each test. For the test, you code must complete execution within the proscribed time limit (10 seconds) or it will receive no credit. The tester can be run from the shell or by using a testing framework in an IDE (like Nose or Py.test). Be sure that your code works correctly with the provided, unmodified testing suite.

Submitting your Assignment
You should complete the “pressure_pd_solution”, “rocket_pid_solution”,

“bipropellant_rocket_pid_solution” functions in the RocketPIDStudent_submission.py file to satisfy the specifications described therein

(by implementing a PID controller), and then you should submit the file to the Rocket PID Assignment on Canvas - Gradescope. Do not place the file into an archive (zip, tar, etc.), and only submit RocketPIDStudent_submission.py without changing the file name. You may submit any number of times before the deadline. Your submission should use Python version 3 and may optionally use external modules (Matplotlib, NumPy, SciPy, etc.) that are present in the respective evaluation files. Your python file must execute NO code when it is imported. We encourage you to keep any testing code in a separate file that you do not submit. They should also NOT display a GUI or Visualization when we import them or call your function under test. If submissions do not follow naming convention or we have to manually edit your code to comment out your own testing harness or visualization you will receive a -20 point penalty.

Grading
Grading for Part A is based on how well your controller is capable of maintaining a stable pump pressure. If the pump’s pressure drops below 0 or climbs above 105 during any part of the test, it will result in 0 points. Otherwise your score will be calculated based on how closely the pump’s pressure is maintained at 100 for the duration of the test. Part A is worth 25% of the grade.

The grading for Part B is broken into two parts: * maintaining two specific velocity stages * avoiding excessive speed on landing

If your PID controller fails to produce valid throttle control, you may receive zero credit. If your PID controller is able to stay on course for 130 seconds during the simulated flight, you will receive a score of 65 which is full credit for that section of part B. Partial credit will be awarded for less than desirable flight control. In the landing section, your controller must make a controlled descent to land under the acceptable velocity for which you will receive the landing score of 35. You will not receive partial credit for a “crash” landing. These two scores will add up to your final score for part B with a maximum possible score of 100 which denotes full credit for Part B. Part B is worth 65% of the total grade for the PID project.

The grading for Part C is broken into two parts: * maintaining the upward specific velocity stage * avoiding excessive speed on landing

If your PID controller fails to produce valid throttle control, you may receive zero credit. If your PID controller is able to stay on course for 130 seconds during the simulated flight, you will receive a score of 65 which is full credit for that section of part C. Partial credit will be awarded for less than desirable flight control. In the landing section, your controller must make a controlled descent to land under the acceptable velocity for which you will receive the landing score of 35. You will not receive partial credit for a “crash” landing. These two scores will add up to your final score for part C with a maximum possible score of 100 which denotes full credit for Part C. Part C is worth 10% of the total grade for the PID project.

Note that the grader also includes an assert true warning that you may ignore, but it is present to alert you that all conditions of the solution have not been met. Also, the velocity profile provided is simply a guideline and you may use whatever system to define throttling you desire. Final grading will use the same controller parameters as in the student submission.

Academic Integrity
You must write the code for this project alone. While you may make limited usage of outside resources, keep in mind that you must cite any such resources you use in your work (for example, you should use comments to denote a snippet of code obtained from StackOverflow, lecture videos, etc)