Project: Motion Planning and Decision Making for Autonomous Vehicles

In this project, I have implemented two of the main components of a traditional hierarchical planner: The Behavior Planner and the Motion Planner. Both will work in unison to be able to:

1. Avoid static objects (cars, bicycles and trucks) parked on the side of the road (but still invading the lane). The vehicle must avoid crashing with these vehicles by executing either a “nudge” or a “lane change” maneuver.

2. Handle any type of intersection (3-way, 4-way intersections and roundabouts) by STOPPING in all of them (by default)

3. Track the centerline on the traveling lane.

To accomplish this, I have implemented:

• Behavioral planning logic using Finite State Machines - FSM
• Static objects collision checking.
• Path and trajectory generation using cubic spirals
• Best trajectory selection though a cost function evaluation. This cost function will mainly perform a collision check and a proximity check to bring cost higher as we get closer or collide with objects but maintaining a bias to stay closer to the lane center line.

Dependency - Udacity VM Workspace

The state code in this repository is aligned to run on the Udacity VM workspace. Refer to the classroom page Ubuntu VM Workspace - Overview to learn how to access the VM workspace and its restrictions and best practices.

However, to set up your local machine with the necessary tools, you must have either Windows Subsystem for Linux (WSL) or Ubuntu 20.04 or 18.04 LTS. Below is the list of tools installed in the Udacity VM workspace that you should install on your local machine.

• CARLA simulator 0.9.9.4.
  You can find more details at CARLA Quick Start Installation. The deb installation is the easiest way to get the latest release in Linux.

  sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 1AF1527DE64CB8D9
  sudo add-apt-repository "deb [arch=amd64] http://dist.carla.org/carla $(lsb_release -sc) main"
  sudo apt-get update # Update the Debian package index
  sudo apt-get install carla-simulator=0.9.10-2 

  The installation directory must be /opt/carla-simulator/ on your Linux machine. To verify, open a terminal an launch CARLA as:

  cd /opt/carla-simulator
  ./CarlaUE4.sh

  The Carla Simulator should launch in a few seconds. You can close it after verification.

• NICE DCV Server.
  This includes the Nvidia drivers along with CUDA libraries for the underlying Tesla T4 GPU.

  Sat Oct 14 15:31:45 2023       
  +---------------------------------------------------------------------------------------+
  | NVIDIA-SMI 535.104.12             Driver Version: 535.104.12   CUDA Version: 12.2     |
  |-----------------------------------------+----------------------+----------------------+
  | GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
  | Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
  |                                         |                      |               MIG M. |
  |=========================================+======================+======================|
  |   0  Tesla T4                       On  | 00000000:00:1E.0 Off |                    0 |
  | N/A   31C    P0              27W /  70W |   2093MiB / 15360MiB |     27%      Default |
  |                                         |                      |                  N/A |
  +-----------------------------------------+----------------------+----------------------+
                                                                                          
  +---------------------------------------------------------------------------------------+
  | Processes:                                                                            |
  |  GPU   GI   CI        PID   Type   Process name                            GPU Memory |
  |        ID   ID                                                             Usage      |
  |=======================================================================================|
  |    0   N/A  N/A      1055      G   /usr/lib/xorg/Xorg                           67MiB |
  |    0   N/A  N/A      1521      G   /usr/lib/xorg/Xorg                           89MiB |
  |    0   N/A  N/A      1669      G   /usr/bin/gnome-shell                         23MiB |
  |    0   N/A  N/A      1948    C+G   /usr/lib/x86_64-linux-gnu/dcv/dcvagent      398MiB |
  |    0   N/A  N/A      3320      G   ...sion,SpareRendererForSitePerProcess       30MiB |
  |    0   N/A  N/A      4489    C+G   ...aries/Linux/CarlaUE4-Linux-Shipping     1348MiB |
  +---------------------------------------------------------------------------------------+

  dcv version
  # Output
  NICE DCV 2023.0 (r15487)
  Copyright (C) 2010-2023 NICE s.r.l.

• C++

  gcc --version
  # Output
  gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0

• Git

• OpenCV

• CMake and Make

• VSCode

• Eigen Library for C++

• Point Cloud Library

• Python3 and Pip v20.xx or above.

  python3 --version
  # Output
  Python 3.8.10

• ROS

• Project specific dependencies

  # Required for building project
  sudo apt-get install -y libgoogle-glog-dev libgtest-dev
  # Required for running project. 
  # Install carla python package
  sudo python3 /usr/lib/python3/dist-packages/easy_install.py /opt/carla-simulator/PythonAPI/carla/dist/carla-0.9.9-py3.7-linux-x86_64.egg
  # Install python requirements
  pip install numpy pygame websocket-client

Instructions

The sections ahead will guide you through the steps to build and run the project.

Step 1. Log into VM Workspace

Open the VM workspace and log into the VM to practice the current project. Once you log into the VM, open a Terminal window.

Step 2. Clone the Repository

Fork the repository to your Github account and clone the repository to the workspace using the commands below.

git clone https://github.com/udacity/nd013-c5-planning-starter.git

Change to the project directory.

cd nd013-c5-planning-starter/project

Step 3. Review the starter files

You will find the following files in the project directory.

.
├── cserver_dir
├── install-ubuntu.sh
├── manual_control.py
├── run_carla.sh
├── run_main.sh
├── simulatorAPI.py
└── starter_files

Step 4. Start the Carla Server

Start the Carla server by executing the following shell script.

./run_carla.sh

This file will, in turn, execute the /opt/carla-simulator/CarlaUE4.sh script.

Step 5. Install Dependencies

Open another Terminal tab, and change to the nd013-c5-planning-starter/project directory. Execute the following shell script to install the project-specific dependencies.

./install-ubuntu.sh

This file will install utilities such as, libuv1-dev, libssl-dev, libz-dev, uWebSockets.

Step 6. Update the Project Code

Change to the starter_files/ directory

cd starter_files/

Before you start coding, we strongly recommend you look at the rubric in your classroom, against which the human Mentor will review your submission. Your submission must satisfy all rubric criteria to pass the project; otherwise, the Mentor may ask you to re-submit.

Update the following files as per the classroom instructions.

• behavior_planner_FSM.cpp
• cost_functions.cpp
• motion_planner.cpp
• velocity_profile_generator.cpp
• planning_params.h

Important: At this moment, it is important to save your work and push it back to the remote Github repository.

Update Notes

In the previous version of the project starter code, we had libcarla-install/ and rpclib/ directories inside the starter files. But, those directories are no longer needed in the current version of the starter code because the current CMakeLists.txt file has corresponding includes and libs added at /opt/carla-source.

To give some old context, when we had rpclib/ directory inside the starter files, we used to compile the rpclib library using the following commands.

cd starter_files
rm -rf rpclib
git clone https://github.com/rpclib/rpclib.git

This library is a msgpack-rpc library written using modern C++. The goal of building this library was to provide a simple RPC solution. However, all of the above-mentioned steps are no longer needed in the current version of the project strarter code.

Step 7. Build and Execute the Project

When you finish updating the project files, you can execute the project using the commands below.

# Build the project
# Run the following commands from the starter_files/ directory
cmake .
# The command below compiles your c++ code. Run it after each time you edit the CPP or Header files
make

# Run the project
cd ..
# Run the following commands from the nd013-c5-planning-starter/project directory
./run_main.sh

If the execution fails silently, you can use ctrl + C to stop, and try again.

Another possible error you may get is bind failed. Error: Address already in use. In such a case, you can kill the process occupying the required port using the commands below.

ps -aux | grep carla
# Use the IDs displayed in the output of the last command. 
kill id     

Note:

This project is now a completed portfolio piece, showcasing skills in motion planning and decision-making for autonomous vehicles. It is part of the Udacity Self-Driving Cars Nanodegree.

Ensure that the project meets all rubric criteria before submitting it to your portfolio or showcasing it to potential recruiters.