Autonomous Vehicle Dynamics and Control MSc

The aim of this module is to provide the AVDC MSc students with an overview of the course and to introduce the main aspects the Autonomous Systems underpinning the course, including interactive flight demonstrations of UAS. This module structure also allows it to be offered as a short course under the same title.

• Introduction to the AVDC course
• UAS design parameters
• UAS composite materials & structures
• UAS: Strength of structures
• UAS flight demonstration: Indoor flight
• UAS components: mechanical & electrical
• Overview of UAS Power & Propulsion
• Overview of UAS Guidance & Navigation
• Overview of UAS Sensor Fusion
• Overview of Artificial Intelligence for UAS
• Overview of UAS Operations
• UAS flight demonstration: Outdoor flight

On successful completion of this module a student should be able to:

1) Specify the main practical applications of Unmanned Aircraft Systems (UAS) and define their engineering subsystems.
2) Evaluate the main engineering challenges of UAS analysis and design.
3) Analyse qualitatively the functions and capabilities of the main subsystems of UAS
4) Specify ethical concerns and legal issues concerning UAS operations

Mathematical modelling and simulation of unmanned aerial vehicles is a vital part of system development. Nowadays COTS components becoming more powerful and can multi-task and carry-out complex computations. The student would need to learn the technical skills not only for the modelling and simulation but also the real-time implementation of the algorithms. The aims of this course are to provide the student with the skills and knowledge necessary to model, simulate and then critically analyse the resultant non-linear motion of unmanned air vehicles using mainly Matlab/Simulink and target compile the algorithms on an embedded flight control system.

• Introduction to mathematical modelling and simulation; systems of nonlinear ODEs; equilibrium, linearisation and stability; numerical & computational tools (10 hours).
• Model building; model testing, validation and management; trimming and numerical linearisation (10 hours).
• Control implementation and testing on COTS FCS

On successful completion of this module a student should be able to:

1. Evaluate and implement an example UAV in terms of their aerodynamic, control, mass and inertia characteristics. Appraise and critically compare the resultant motion.
2. Distinguish the requirements for model testing, verification and validation, and demonstrate their application to an UAV model.
3. Implement and apply selected control laws on a commercial of the shelve flight control board and carry out simulation-in-the-loop testing.
4. Communicate and present both orally and in writing results of individual work.

The aim of this module is to provide an overview of sensor fusion architectures, algorithms and applications in the context of autonomous vehicles navigation, guidance and control both for linear and non-linear systems. The module aims also to give the students an understanding of the appropriate tools for error analysis, diagnostic statistics and heuristics enabling them to critically evaluate the performance of a sensor fusion architecture/algorithm. The main emphasis is on the Kalman Filter algorithm together with variants and generalisations, applied to target tracking problems.

• Statistical Analysis (4 lectures)
• Linear Kalman Filter and Linear Kalman Smoother (5 lectures)
• Inertial navigation (3 lectures)
• Constrained filters (1 lecture)
• Sensor Integration architectures and Multiple sensor fusion (3 lectures)
• Non-linear filters (EKF, UKF and Particle Filters) (5 lectures)
• Case Study: Inertial navigation (3 lectures)
• Case Study: Multiple sensor fusion (3 lectures)

On successful completion of this module a student should be able to:

1. Understand the fundamental principles in stochastic processes and in estimation theory.
2. Formulate, set up and execute the Kalman filter to linear processes and be able to assess the functional operation of the filter.
3. Formulate, set up and execute non-linear filters (Extended Kalman filter, Unscented Kalman Filter, Particle filters) to non-linear or non-Gaussian models.
4. List common motion models used in target tracking and navigation applications.
5. Design and appraise the performance of multi-sensor fusion architectures in a real-case scenario.

The aim of this module is to introduce the AVDC MSc students to the Artificial Intelligence algorithms suitable for real life problems concerning the Autonomous Systems (AS): target detection, identification, recognition and tracking using multiple heterogeneous sensors from cooperating AS, including accuracy assessment and uncertainty reduction for these applications.

• Introduction to AI for AS (1 lecture)
• AS sensors and imaging (4 lectures)
• Demo: Indoor autonomous surveillance (2 lectures)
• AI Algorithms: Supervised Learning―Neural Networks (7 lectures)
• AI Algorithms: Supervised Learning―Lab (3 lectures)
• AI Algorithms: Unsupervised Learning―Clustering (3 lectures)
• Automated Reasoning: Bayesian Networks (2 lectures)
• Relevance Filtering: Visual Attention (2 lectures)
• Case Study: AI for AS (4 lectures)

On successful completion of this module a student should be able to:

1) Formulate the fundamental meaning and applicability of Artificial Intelligence (AI) algorithms for Autonomous Systems (AS).
2) Categorize AI methods for real-life scenarios of AS applications.
3) Set up the commonly used AI algorithms for application in the AS context.
4) Evaluate performance of AI algorithms for a typical AS application in a simulation environment.

This module aims to provide fundamental and critical understanding of classical and advanced guidance and navigation approaches for Autonomous Systems. A key aim is for the students to critically understand the problems of guidance and navigation, the fundamental solution approaches together with their advantages and limitations.

• Introduction on navigation and guidance systems;
• Path planning for autonomous systems
• Path following for autonomous systems
• UAV (Unmanned Aerial Vehicle) guidance systems;
• Guidance approaches: conventional guidance such as PN (Proportional Navigation), geometric guidance, and optimal guidance;
• Navigation approaches: navigation systems, GNSS (Global Navigation Satellite System), terrain based navigation, SLAM (Simultaneous Localisation and Mapping);
• Cooperative guidance and collision avoidance.

On successful completion of this module a student should be able to:

1. Evaluate the fundamentals of the various guidance techniques and their properties.
2. Appraise the algorithms that are required to produce an estimate of position and attitude;
3. Critically examine the characteristics, purposes, and design procedures of guidance and navigation systems;
4. Assess the challenges of guidance and navigation approaches for autonomous systems;
5. Examine the challenges of cooperative guidance and navigation approaches for autonomous systems.

This module aims to provide students with fundamental understanding and knowledge of the advanced control systems and their applications to autonomous vehicles. Building upon the foundational “UAS Flight Control” module, advanced control methods are presented and analysed to mitigate limitations of the conventional linear control approaches. A key aim is for the students to critically understand the properties of the system model, the fundamental working principals of the advanced control approaches together with their advantages and limitations.

• Overview of autonomous vehicles performance requirement;
• System identification and state-space modelling;
• Uncertainty representation;
• Pole placement and gain scheduling;
• H-infinity control: loop shaping approach and LMI approach;
• Optimal and non-linear control system design and its application to autonomous vehicles;
• Gain scheduling

On successful completion of this module a student should be able to:

1. Appraise the nature, purpose, design procedure of advanced control systems for autonomous vehicles;
2. Evaluate the limitations of conventional control and their impact on autonomous vehicles;
3. Critically analyse adaptive control theories, their properties, and applications to autonomous vehicles;
4. Design a robust controller for autonomous aerial vehicles and critically evaluate the performance of the robust control system;
5. Analyse the stability, robustness and sensitivity of the control systems for autonomous vehicles.

• Dr Marta Ceccaroni

This module aims at providing the foundations of logic and reasoning modelling to understand how reasoning can be automated. Moreover it introduces the fundamental techniques and formal language to design automated reasoners.

• Introduction to logical representation and reasoning
• Logical Agents
• Propositional Logic
• First-order Logic
• Inference
• Engineering domain knowledge representation
• Exercises and case studies

On successful completion of this module a student should be able to:

1.    Analyse syntax and semantics of first-order logic and different knowledge representation using the appropriate formal language.

2.    Use logical programming to build knowledge bases.

3.    Design strategies to solve assigned reasoning problems detailing the execution steps to perform the given tasks.