85804 - Neural Network Computing, A.I. and Machine Learning for Automotive M

Learning outcomes

The course will introduce the basic aspects of modern artificial intelligence for reasoning on data, and the software for learning by data, with a specific use of Neural Networks. In particular it will discuss: - Fundamentals of AI : giving examples and models of computer-based solutions that are capable of intelligent behavior, using a priori knowledge, Sensing, Perception, Knowledge, Reasoning and Learning. - Fundamentals of Machine Learning: developing on tools and systems that can learn from and make predictions on data, for getting machines to act without being explicitly programmed. - Fundamentals of Deep Learning: for modeling and implementing deep neural network architectures and algorithms. The course will be based on laboratory for applications in automotive fields such as in autonomous driving and in automatic classification of external sensory data. Specific topics selected by companies will be considered for lab projects.

Course contents

# History and impact of AI in Automotive
Introduction to the history of AI and to its economical impact in the Automotive domain. AI and Computer Vision for Automotive.

# Neural Architectures fundamentals
Gradient-based optimization, fully connected and convolutional architectures. Analysis of computational costs and computational graph management. Design practices, network surgery.

# On-board sensors and cameras, depth sensors
Analysis of commercial on-board sensors, depth, thermal and RGB cameras.

# Sequence modelling and prediction
Recurrent, Convolutional and Fully-Attentive architectures for sequence understanding and generation. Computational graph analysis. Self-attention and cross-attention. Applications to trajectory prediction and language-based interfaces.

# Visual understanding algorithms
Algorithms and techniques for motion estimation, architectures for video object detection and trajectory prediction. Planar Distance estimation, road segmentation, road lane detection. Driver Monitoring, Driver Distraction prediction, eye fixation prediction.

# Image generation
Generative algorithms for images and videos: Generative Adversarial Networks, VAEs; techniques for rendering high-resolution images. Applications to the automotive domain.

# Reinforcement learning for navigation
Markov decision processes, policy learning strategies. Algorithms for locomotion and navigation. Simulated environments and strategies for deployment in real settings. Applications to autonomous driving.

# Architecture optimization and compression
Weight matrix factorization, quantization and pruning approaches. Design practices for optimization.

Readings/Bibliography

Slides and scientific papers from international conferences and journals (CVPR; ECCV; ICCV; T-PAMi. IVPR, IEEE ITS; IEEE IV)

Teaching methods

Most of the lessons are frontal and use scientific slides and papers as didactic support; about 30% of the lessons are laboratory, with hands-on experiences on tensor computation libraries and the design and training of neural networks and AI algorithms for the Automotive. Upon completion, meetings and discussions with companies are organized. Students are required to develop a final project (to be presented during the exam) which requires the study and development of a predictive algorithm applicable in the automotive field, its experimental validation and possibly the collection of data. The presentation of the project must be accompanied by the delivery of the developed code and a technical report describing the approach, its positioning in the literature, the datasets used and the experimental results obtained. All lessons will be streamed on Microsoft Teams. The recordings of the lessons will also be available on the Dolly page of the course.

Assessment methods

The exam consists of an oral test, with theoretical questions, presentation of the laboratories developed during the course, and presentation of the final project (see the section "teaching methods").

Office hours

See the website of Rita Cucchiara

See the website of Lorenzo Baraldi