/CarND-MPC-Project

CarND Term 2 Model Predictive Control (MPC) Project

Primary LanguageC++MIT LicenseMIT

CarND-Controls-MPC

Self-Driving Car Engineer Nanodegree Program

Abstract

This repository contains the last project of term2. The goal of this project is to navigate a track in simulator, concretely, sending steering and acceleration commands to simulator and let car finish the track with less error(cross track error and orientation).in addition, there is a time latency in real world from getting data to make actuations. The solution should be robust for it.

The solution makes use of IPOPT and CPPAD libraries to calculate an optimal trajectory and related actuations to minimize error with a third-degree polynomial fit to the given waypoints. we should let the MPC only considers a short duration, produces a trajectory for that duration based on a model of vehicle's kinematics and a cost function based mostly on vehicle's cross track error and orientation error, and other factors that may improve performance.

Rubric Points

The Model

The model contains a 6 parameters' state(xt, yt, psit, vt, ctet, epsit) as input and expects a output (delta, a). The actuators are limited, which contains steering angle(+-25degrees) and throttle(+-1,+1 denotes full throttle and -1 means full breaking).Based on the equations in previous course

  xt+1 = (xt + vt * cos(psit) * dt);
  yt+1 = (yt + vt * sin(psit) * dt);
  psit+1 = (psit - vt * deltat / Lf * dt);
  vt+1 = (vt + at * dt);
  ctet+1 = ((ft - yt) + (vt * sin(epsit) * dt));
  epsit+1 - ((psit - psidest) - vt * deltat / Lf * dt);

Timestep Length and Elapsed Duration (N & dt)

N means the total number of points we are considering to calculate the future state of the vehicle. dt is the time interval between the two continuous states. In this implementation, I finally use N=10 and dt=0.1 to track a range of T=1s.

How to tune the parameters(N and dt)?

Firstly, I use the values same as what I did in course, which is 25 in N and 0.05 in dt. But as what I imagined that the vehicle just rushed to road at begin. In my mine, I think df will reflect the real state if it has a small value, and the nearer the prediction to current state, the more correct output we will get. In addition, larger value of N will cost a lot of computation and we just use some of it, the rest will dropped as we will implement next prediction.

How does T(N * dt) affect the prediction?

Maybe increasing N and decreasing dt is a good choice to increase the accuracy of prediction. But if the car at high speed, we have large N and very small dt, then the computation will take a lot of time that the vehicle can't have actuation in time.

Polynomial Fitting and MPC Preprocessing

Compare to MPC quiz, this project we need convert the global coordinates ptsx and ptsy to vehicle's local coordinate waypoints_x and waypoints_y. Then I use polyfit on waypoints to get coefficients of 3rd-degree curve in the line the car will be moving. Based on these coefficients I can compute the cross track error and orientation error, both of them will be used to next prediction.

Model Predictive Control with Latency

In my implement, I take the latency into account that I use it to compute a future state as my current state, then take it in my MPC model to predict, which you can find at line 110-122 in main.cpp

Dependencies

Basic Build Instructions

  1. Clone this repo.
  2. Make a build directory: mkdir build && cd build
  3. Compile: cmake .. && make
  4. Run it: ./mpc.

Tips

  1. It's recommended to test the MPC on basic examples to see if your implementation behaves as desired. One possible example is the vehicle starting offset of a straight line (reference). If the MPC implementation is correct, after some number of timesteps (not too many) it should find and track the reference line.
  2. The lake_track_waypoints.csv file has the waypoints of the lake track. You could use this to fit polynomials and points and see of how well your model tracks curve. NOTE: This file might be not completely in sync with the simulator so your solution should NOT depend on it.
  3. For visualization this C++ matplotlib wrapper could be helpful.)
  4. Tips for setting up your environment are available here
  5. VM Latency: Some students have reported differences in behavior using VM's ostensibly a result of latency. Please let us know if issues arise as a result of a VM environment.

Editor Settings

We've purposefully kept editor configuration files out of this repo in order to keep it as simple and environment agnostic as possible. However, we recommend using the following settings:

  • indent using spaces
  • set tab width to 2 spaces (keeps the matrices in source code aligned)

Code Style

Please (do your best to) stick to Google's C++ style guide.

Project Instructions and Rubric

Note: regardless of the changes you make, your project must be buildable using cmake and make!

More information is only accessible by people who are already enrolled in Term 2 of CarND. If you are enrolled, see the project page for instructions and the project rubric.

Hints!

  • You don't have to follow this directory structure, but if you do, your work will span all of the .cpp files here. Keep an eye out for TODOs.

Call for IDE Profiles Pull Requests

Help your fellow students!

We decided to create Makefiles with cmake to keep this project as platform agnostic as possible. Similarly, we omitted IDE profiles in order to we ensure that students don't feel pressured to use one IDE or another.

However! I'd love to help people get up and running with their IDEs of choice. If you've created a profile for an IDE that you think other students would appreciate, we'd love to have you add the requisite profile files and instructions to ide_profiles/. For example if you wanted to add a VS Code profile, you'd add:

  • /ide_profiles/vscode/.vscode
  • /ide_profiles/vscode/README.md

The README should explain what the profile does, how to take advantage of it, and how to install it.

Frankly, I've never been involved in a project with multiple IDE profiles before. I believe the best way to handle this would be to keep them out of the repo root to avoid clutter. My expectation is that most profiles will include instructions to copy files to a new location to get picked up by the IDE, but that's just a guess.

One last note here: regardless of the IDE used, every submitted project must still be compilable with cmake and make./

How to write a README

A well written README file can enhance your project and portfolio. Develop your abilities to create professional README files by completing this free course.