This is an old revision of the document!
Module: Control, Planning, and Decision-Making (Part 2)
| Study level | Master | |
| ECTS credits | 1 ECTS | |
| Study forms | Hybrid or fully online | |
| Module aims | To provide students with advanced theoretical and practical knowledge in the validation and verification of control, planning, and decision-making systems used in autonomous platforms. The module explores simulation-based testing, formal verification, and model-checking techniques that ensure the safety and stability of AI-enhanced controllers. Students will learn to integrate hybrid simulation frameworks, formal reasoning tools, and optimization algorithms to assess real-world readiness of control architectures. Emphasis is placed on quantifiable safety guarantees, robustness against uncertainty, and compliance with international standards (ISO 26262, ISO 21448, and IEEE 2846). | |
| Pre-requirements | Strong background in control theory, optimization, and planning algorithms. Familiarity with programming (Python, C++, MATLAB), model-based design tools (Simulink, ROS 2), and AI decision-making frameworks. Basic understanding of formal methods, hybrid systems, and system modeling. Prior exposure to simulation environments or real-time control applications is recommended. | |
| Learning outcomes | Knowledge: • Explain simulation-based and formal validation approaches for control and planning systems. • Describe the use of model-checking, reachability analysis, and verification frameworks in autonomous systems. • Understand ISO 26262, ISO 21448 (SOTIF), and IEEE 2846 standards relevant to control and decision-making validation. • Discuss trade-offs between simulation fidelity, computational efficiency, and real-time constraints. Skills: • Develop and validate control and planning algorithms in simulation environments (MATLAB/Simulink, ROS 2, CARLA). • Apply formal verification tools (UPPAAL, SPIN, or CBMC) to analyze safety and correctness properties. • Design hybrid validation workflows combining Monte Carlo simulation and symbolic reasoning. • Evaluate algorithm robustness and decision safety under stochastic and adversarial conditions. Understanding/Attitudes: • Appreciate the role of rigorous validation in certifying autonomous behaviors and AI-based decision-making. • Recognize limitations of current simulation and formal verification tools in high-dimensional, data-driven systems. • Adopt ethical, transparent, and standards-compliant practices in the assurance of autonomy. |
|
| Topics | 1. Validation of Control and Planning Systems: – System-level validation frameworks and verification-driven design. – Simulation fidelity, corner-case testing, and scenario coverage. 2. Simulation Environments and Tools: – SIL/HIL setups, Monte Carlo analysis, and statistical validation. – Multi-domain co-simulation for cyber-physical systems. 3. Formal Verification and Model Checking: – Safety property specification and temporal logic (LTL, CTL). – Reachability analysis, invariant verification, and constraint solving. 4. Hybrid and Nonlinear Systems: – Modeling hybrid automata and nonlinear control loops. – Formal abstraction and conservative over-approximation techniques. 5. Standards and Safety Frameworks: – ISO 26262, ISO 21448, IEEE 2846, and ASAM OpenSCENARIO for validation. 6. Case Studies: – Autonomous driving, UAV flight control, and robotic path planning validation. |
|
| Type of assessment | The prerequisite of a positive grade is a positive evaluation of module topics and presentation of practical work results with required documentation | |
| Learning methods | Lectures: Cover theory and methodologies for simulation-based and formal validation of control and planning systems. Lab works: Implement and test controllers in virtual and hybrid environments (ROS 2, MATLAB, CARLA, Scenic, CommonRoad, UPPAAL). Individual assignments: Develop validation pipelines, perform reachability analysis, and document results. Self-learning: Study research papers and international standards on autonomy verification and formal safety assurance. |
|
| AI involvement | Yes — AI tools may be used to automate scenario generation, identify unsafe trajectories, and optimize validation coverage. Students must validate AI-assisted outcomes, ensure reproducibility, and cite AI involvement transparently in deliverables. | |
| References to literature | 1. Rajamani, R. (2012). Vehicle Dynamics and Control (2nd ed.). Springer. 2. Baier, C., & Katoen, J.-P. (2018). Principles of Model Checking. MIT Press. 3. Koopman, P., & Widen, J. (2023). The AI Driver: Defining Human-Equivalent Safety for Automated Vehicles. IEEE Transactions on Intelligent Vehicles. 4. ISO 21448 (2022). Road Vehicles – Safety Of The Intended Functionality (SOTIF). 5. ISO 26262 (2018). Road Vehicles – Functional Safety. 6. IEEE 2846 (2022). Assumptions for Models in Safety-Related Automated Vehicle Behavior. 7. Althoff, M., et al. (2021). Formal Verification of Autonomous Systems: State of the Art and Future Directions. IEEE Access. 8. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. |
|
| Lab equipment | Yes | |
| Virtual lab | Yes | |
| MOOC course | Suggested MOOC: 'Formal Methods for Autonomous Systems' (edX, University of York) or 'Verification and Validation of AI Systems' (Coursera, Stanford University). | |
