Perception and decision making for intelligent robot navigation

This repository contains the material of the Design Project, with group nr 5 and title "Perception and decision making for intelligent robot navigation (Navi_Perception)"

Background

Autonomous Robots are expected to be deployed in offices, factories or household scenarios where the robot will share spaces with humans. For this, one important capability of such autonomous robots is to have a robust decision-making system that is able to consider dynamic environments, such as obstacles. In order to satisfy the time and place requirements, the robot should have a robust navigation system allowing it to negotiate the different obstacles in its path to achieve its desired destination. The dynamic and stochastic nature of office environments requires a proper method to fuse the different sensors of a robot. For example, the robot should learn and decide in run-time which sensor to use depending on the perceived context information. This will allow robots to react efficiently to potential collisions while keeping as much as possible the desired goal. However, the complex environment that a robot can encounter produces noise in the sensors, e.g., odometry errors and uncertainty in the actuators (wheel velocities). The intelligent navigation system should efficiently decide when to switch to the sensor that gives better information about the environment to navigate safely.

Students:

• Yaochen Song yaochen@chalmers.se

Advisors:

• Karinne Ramirez-Amaro karinne@chalmers.se
• Stephanie Berrio

Problem and Goal

The goal of this project is to investigate how different robotic sensors are used for autonomous navigation in indoor environments.

For the final report, pls check "Documentation/final_handin" in main branch

The video link is shown below:

https://www.youtube.com/playlist?list=PLfi-srdJh5ddZDsl7bJtQfmNswJQYdw1R

Git repo structure

In general, the solutions generated within this project will be integrated within a larger scenario. Therefore, it is important to maintain the modularity and reusability of the obtained solutions.

In general, this project can produce source code (C++, Python, and ROS), diagrams, designs, and documentation that may be re-used and integrated with other projects. Most probably, you will need to implement additional code, routines, or tests to develop your solutions. For example, if you are training a NN, you will need to pre-process and label the training data. This is material needed to develop your solution but it is not part of the solution. Therefore, it should not be included directly in the resultant material.

For this reason, we propose the following file system for this repo:

.
├── Code (Folder that contains all the code generated in this project)
│   └── python_tools (Folder that contain all the python code and additional tools to support your development)
|   └── ros (Folder containing ros workspaces and/or ros packages). NOTE: Do not include binary and log folders, e.g. build/ devel/ log/
├── Diagrams (This folder contains circuit diagrams, e.g., schematics (.sch), boards (.brd), CAD (.stl, .cda), etc. generated and used in this project)
├── Documentation (This folder contains all the documentation generated and used in this project)
│   ├── How2s (Folder with mini-how-tos providing instructions on how to run demos or applications developed within this project)
│   ├── PlanningReport (Folder containing material for the planning report)
│   ├── Resources (Folder containing material from third party authors, e.g. reference papers (PDF), figures, manuals, etc.)
│   ├── Thesis/Report (Folder containing the final report or thesis)
│   └── Workdiary (Folder with a work diary template to report the daily activities)
└── README.md (Main project's readme file, with these general instructions and documentation's links to understand this project.)

You can add more folders to improve the file system but the general structure should be preserved. The idea is to provide a homogeneous structure that is preserved through all the connected projects.

Getting started

Agile Project Management (APM) with Gitlab

We are planning to use APM to manage the project (https://kanbanize.com/agile/project-management). ​

We have prepared a set of videos that explain how to use Github to implement APM. Please, follow the videos.

https://www.youtube.com/playlist?list=PLObANLLE5pkmDIPhaBsUbmjp01qI8KxvU

Ubuntu tutorials

The Linux command line for beginners: https://ubuntu.com/tutorials/command-line-for-beginners#1-overview

Ros version: ROS Noetic

Ubuntu install of ROS Noetic: http://wiki.ros.org/noetic/Installation/Ubuntu

Ros tutorials

http://wiki.ros.org/ROS/Tutorials (general ros tutorials)

https://gitlab.com/deanch/ros_course_template (Dean's tutorial, more specific)

Wiki Material

More information will be provided in the wiki pages.

Code Standards and Formats

ROS style and naming conventions

ROS provides a comprehensive guide on naming conventions and coding standards.

You can find the guidelines for python here:

http://wiki.ros.org/PyStyleGuide

And for C++, here

http://wiki.ros.org/CppStyleGuide

The guidelines for naming convention and units can be found here:

http://wiki.ros.org/ROS/Patterns/Conventions

Finally, here are some instructions on how to format C++ code using ROS format

https://github.com/PickNikRobotics/roscpp_code_format

Python

We would appreciate if you write the code in this repo in accordance with the following standards:

• black formatting https://black.readthedocs.io/en/stable/
• pylint https://pylint.org/
• pydocstyle http://www.pydocstyle.org/en/stable/

Please use the standard settings for these packages. They are easily installed with pip (preferable on a conda virtual environment):

• pip install black
• pip install pylint
• pip install pydocstyle.

Once installed, you can check your python scripts (e.g. test.py) through

• black test.py (add flags --check --diff to check the file without making any changes)
• pylint test.py
• pydocstyle test.py

These commands also work on entire folders. However, please get familiar with these tools before reformatting large amounts of code!

Furthermore, these tools work well in VSCode (surely in other IDEs as well).

How to activate black formatting in VSCode

https://dev.to/adamlombard/how-to-use-the-black-python-code-formatter-in-vscode-3lo0

How to activate linting (e.g. pylint) in VSCode

https://code.visualstudio.com/docs/python/linting

Make sure to also visit settings (through ctrl+,) and search for "python.linting.pylintArgs" and add "--enable=F,E,W" as an item. Otherwise any pylint warnings will be suppressed.

How to activate pydocstyle in VSCode

Open settings with ctrl+, and search for "python.linting.pydocstyleEnabled". Enable for both user and workspace.