Lecture notes and reading material
Preliminary lecture notes will be available in PDF format before the class (usually Monday evening) while definitive lecture notes will be available only after the class has been held, in a timely fashion (usually, at latest a couple of days after the lecture).
Lecture notes will be complemented by possible reading material listed on the syllabus and further pointers, all available on the student area. Due to copyright issues, electronic copies of the material are only available to EPFL students officially enrolled in this course. Students interested in downloading this material can do so from the student area by logging in using with their GASPAR account.
Week 1
TOPIC
Course organization (credits, workload, logistics) and content overview. Introduction to Swarm Intelligence (SI) and key principles, natural and artificial examples. Foraging, trail laying/following mechanisms. Open-space, multi-source foraging experiments: biological data and microscopic models. From real to virtual ants: Ant System (AS), the first combinatorial optimization algorithm based on ant trail/following principles. Application to a classical operational research problem: the Traveling Salesperson Problem (TSP).
LECTURE
Week 2
TOPIC
From AS to Ant Colony Optimization (ACO). Ant-based algorithms (ABC, Ant-Net) applied to routing in telecommunication networks.
LECTURE
Week 3
TOPIC
Introduction to mobile robotics: basic concepts centered around the differential drive vehicle used in the course (e-puck) and the high-fidelity, open-source robotic simulator (Webots). Introduction to control architecture for mobile robots with special focus on reactive control architectures.
LECTURE
Week 4
TOPIC
Localization methods in mobile robotics: positioning systems, odometry-based and feature-based localization. Sources of localization uncertainties and corresponding handling methods for mobile robots.
LECTURE
Week 5
TOPIC
Collective movements in natural societies; focus on flocking phenomena. Collective movements in artificial systems: Reynolds’ virtual agents (Boids) and behavior-based algorithms for flocking and formation in multi-robot systems.
LECTURE
Week 6
TOPIC
Graph-based distributed control for continuous consensus algorithms (spatial rendezvous, formation).
LECTURE
Week 7
TOPIC
Introduction to multi-level modeling techniques (underlying methodological framework, levels, assumptions, principles). Linear and nonlinear modeling case studies.
LECTURE
Week 8
TOPIC
Calibration of model parameters; an additional challenging multi-level modeling case study (distributed seed assembly). Combined modeling and machine-learning methods for control optimization; diversity and specialization metrics.
LECTURE
Week 9
TOPIC
Introduction to unsupervised multi-agent machine-learning techniques for automatic design and optimization: terminology and classification, Particle Swarm Optimization (PSO), performance comparison with Genetic Algorithms. Application of machine-learning techniques to automatic control design and optimization of single-robot systems.
LECTURE
Week 10
TOPIC
Noisy and expensive optimization problems; noise-resistant algorithms. Application of machine-learning techniques to automatic control design and optimization of multi-robot systems. Specific issues for automatic control design and optimization in distributed systems (e.g., credit assignment problem).
LECTURE
Week 11
TOPIC
Division of labor and task-allocation mechanisms: threshold-based algorithms and market-based algorithms.
LECTURE
Week 12
TOPIC
Division of labor and task-allocation mechanisms: comparisons between threshold-based and market-based algorithms. Distributed environmental sensing using static wireless sensor networks.
LECTURE
Week 13
TOPIC
Distributed environmental sensing using static wireless sensor networks.
LECTURE
Week 14
TOPIC
Distributed environmental sensing using robotic sensor networks. General take home messages of the course. Discussion about student feedback for the course.
LECTURE