stc
Course 3: Self-driving Cars & Environment Perception for Navigation in Known and Unknown Space
Workshop Series | Ages: 14+
Total course cost: 60€
14+
120 minutes
Nikos Zourtsanos, Dimitris Piperidis
03/11, 10/11, 17/11, 24/11, 01/12, 08/12, ώρα 15:45 – 17:45
Course 3 falls under the workshop series Self-driving Cars & AI and focuses on the analysis, explanation and synthesis of algorithms for:
- Environment perception from visual data
- Construction of a map of an unknown area
- Finding the optimal route
- Autonomous navigation using a map
Note: Participation in the cycle does not require the completion of the other cycles of the project.
Course 3 consists of the following sessions:
Session 1: General Cycle Description and Introduction to Graph Theory
General description of cycle 3 and a brief introduction to Python programming language. Description of the problem of finding the shortest path and basic introduction to graph theory.
Session 2: Introduction to Neural Networks
Introduction to the artificial neuron (perceptron) and multi-layer neural networks. Methodology for training a neural network. Implementation and execution of a neural network using TensorFlow/Keras libraries.
Session 3: Intersection Recognition with AI
Data collection for recognizing the basic characteristics of the route. Design and implementation of a CNN model for environment perception with final goal of training and applying the model in practice.
Session 4: Algorithms for Finding the Optimal Route
Introduction to basic pathfinding algorithms (DFS, BFS, Dijkstra, and A*). Programming and implementation of pathfinding algorithms.
Session 5: Creating a Map of an Unknown Area
Mapping using intersections. Creation of a graph of the map and data collection via remote control to construct the map.
Session 6: Autonomous Vehicle Navigation Following the Route
Introduction to autonomous vehicle navigation with an AI algorithm and creation of a complete system (map, intersection recognition, optimal route, and control commands). Final experimental testing of the complete system.
References - Supplementary Material
stc