Autonomous Vehicle Systems Final Projects: Policy and List of Possible Topics Objectives
Objectives
The final class projects will be on a topic related to Autonomous or Self-Driving Car technology. The objectives are to study an aspect of the Autonomous Car technologies, perhaps extend or explore the technology and present your findings to the class.
The required submissions are :
(i)_ A Class presentation using Power point slides
(ii) A written report using Word or LaTeX
(iii) A code on the web using GitHub or MATLAB code zip file
Your project report (80 % of the project grade) will be graded according to this list.
• An abstract. (Summary of what you did and your new results). [10 pts.]
• An introduction. (What problem did you try to solve ?). [15 pts.]
• Methods (theory and/or simulations). [15pts.]
• Results (what did you find out or conclude ?). [50pts.]
• References of published related material. [10 pts.]
• Python, MATLAB or “C” code or other software (give a Zip file) or put on GitHub.
Possible Project Topics
MATLAB ADAS Toolbox Autonomous Car Applications Projects
You will be using the MATLAB ADAS Toolbox working on various Demo examples You will need to know about MATLAB, ADAS and other related Toolboxes for the software development.
Apollo v3.0 or v3.5 Autonomous Car Module Applications Projects
You will be using Apollo and working on various Modules of the software.
You will need to know about Apollo, Github and Python and “C” software development.
The project will be to make changes to the module, compile and run Apollo using the Simulators or real time data and show the results of the changes you made to the software.
Here are some Apollo Modules you can work on:
• Routing Module
• Prediction Module
• Perception Module
o Machine Learning Applied to Computer Vision Tasks of Perception
o PSPNet (Pyramid Scene Parsing Network)
• Control Module
• Planning Module
• Etc. Module
CHOOSE ONE OF THE FOLLOWING TOPICS:
Machine Learning Applications Projects
1) Recurrent Neural Networks (RNNs) for Traffic Lane Prediction
2) Machine Learning Applied to Computer Vision Tasks of Perception
a. YOLO: ….. we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast…….
i. https://arxiv.org/abs/1506.02640
ii. https://towardsdatascience.com/yolo-you-only-look-once-real-time-objectdetection-explained-492dc9230006 ELEN 331 Autonomous Vehicle Systems Final Projects Page 3
iii. Original paper (CVPR 2016. OpenCV People’s Choice Award)
https://arxiv.org/pdf/1506.02640v5.pdf
b. YOLOv2 (and YOLO9000)
i.
https://arxiv.org/pdf/1612.08242v1.pdf
c. YOLOv3
i. https://arxiv.org/abs/1804.02767 Other YOLO related resources are at https://pjreddie.com/publications/
3) Machine Learning Applied to Computer Vision Tasks of Perception
a. PSPNet (Pyramid Scene Parsing Network) PSPNet came first in ImageNet scene parsing challenge 2016, PASCAL VOC 2012 benchmark and Cityscapes benchmark. A single PSPNet yields new record of mIoU accuracy 85.4% on PASCAL VOC 2012 and accuracy 80.2% on Cityscapes.
i.
https://arxiv.org/abs/1612.01105
4) Machine Learning Applied to Automomous Vehicle Lane Detection
Apply convolutional neural network to detect road signs or road labels. Try these algorithms on
Databases for Road signs (from Germany), or try Kaggle.
Use Keras with Python for this project instead of OpenCV in C++.
You can use the Dataset here : http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset
And then you will simulate the process on our computer.
Algorithms for Prediction, Planning and Routing
5) Extend A* star algorithm from 2-D to 3-D. Also try out different distances : Euclidean,
Manhattan, Frenet distances.
a. See the weblink https://en.wikipedia.org/wiki/A*_search_algorithm
b. See the weblink https://www.mathworks.com/matlabcentral/fileexchange/26248-a-a-starsearch-
for-path-planning-tutorial
c. See the weblink https://www.geeksforgeeks.org/a-search-algorithm/
6) A study of HD Maps solutions (Study Mercedes-Benz and Tesla approaches)
7) One dimensional Kalman Filters in Python
8) Bayesian Filters with Tracking Dog application in Python
a. See weblink
https://drive.google.com/a/scu.edu/file/d/0By_SW19c1BfhSVFzNHc0SjduNzg/view?usp
=drivesdk