Nemodrive brings together a group of motivated researchers with the required resources and support to build a self-driving car that will be tested on the UPB campus roads. The prototype will be developed to tackle the particularities of campus and other local roads. In this context, some of the recurring challenges of self-driving cars will be empirically put to test, in a small scale, un-controlled environment.

Project description

Given the diversity of perspectives in recent artificial intelligence research, Nemodrive focuses on machine learning techniques. In this project we will investigate one key issue that must be considered when using such algorithms, which is safety and model robustness. The objective is to obtain a realistic evaluation, both quantitatively and qualitatively, of models used in the perception pipeline for driving in the UPB campus (such as object detection and semantic video segmentation). During this project the student will get familiar with most stages of a machine learning pipeline and focus on improving data and evaluation stages.


  • Identify and define procedure for robust analysis of models on local scenario
  • Develop a data augmentation procedure for improving local dataset
  • Evaluate improvement brought by augmentation

Required knowledge

  • Python
  • Enthusiasm and eagerness to learn

Good to know

  • Machine learning basics
  • Neural Networks
  • Convolutional Neural Network
  • Pytorch



Bloomberg Wyliodrin


ROSEdu Logo

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.
Creative Commons License