Apply now
Master'sData, AI, Simulation & Analytics Applications open

Master’s in Reinforcement Learning, Control and Safe AI Engineering

Online master's programme of 1,900 hours in reinforcement learning, control and safe AI engineering. 12 modules covering sequential decision-making, deep RL, model-based learning, classical and modern control, safe RL, verification and functional safety under IEC 61508, ISO 21448 and ISO/IEC TR 5469, with an assigned technical tutor.

Duration1900 h
ModeOnline
LanguageEnglish
Start datesSet by the faculty
AwardSEIUM Institute diploma
Enrol now View syllabus Ask an advisor
Virtual lab · try it before you enrol

Take the controls.

Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.

SIM · LIVE ●
Learning outcomes

What you will be able to do

01

Formulate control problems as Markov decision processes.

02

Implement tabular and deep reinforcement learning algorithms.

03

Compare learned controllers with classical control techniques.

04

Design policies that respect safety constraints.

05

Transfer policies from simulation to the real system.

06

Verify the behaviour of neural-network-based controllers.

07

Build safety arguments in line with functional safety standards.

Profile

Who it's for

  • Reinforcement learning engineers looking to deepen their work on control and safety.
  • Control engineers who want to bring learning-based methods into their designs.
  • Safe AI researchers seeking an applied, standards-based perspective.
  • Autonomous systems engineers responsible for learned decision-making components.
  • Functional safety engineers who need to assess systems that include AI.
  • Decision data scientists moving into sequential decision-making and control.
  • Engineering or science graduates looking to enter robotics.
Standards & tools

What you'll work with

  • IEC 61508 (functional safety of electrical/electronic/programmable electronic systems)
  • ISO 21448 (safety of the intended functionality, SOTIF)
  • ISO/IEC TR 5469 (artificial intelligence and functional safety)
  • ISO 26262 (functional safety of road vehicles)
  • Regulation (EU) 2024/1689 (AI Act)
  • ANSI/UL 4600 (safety evaluation of autonomous products)
  • ISO/IEC 23894 (AI risk management)
  • Gymnasium
  • Stable-Baselines3
  • MATLAB Reinforcement Learning Toolbox
  • MuJoCo
  • Ray RLlib
Syllabus

12 modules · 1900 h

  1. 1.1Markov decision processes
  2. 1.2Reward, return and discount factor
  3. 1.3Policies and value functions
  4. 1.4Bellman equations
  5. 1.5Relationship between optimal control and reinforcement learning
  6. 1.6Partial observability and POMDPs
  7. 1.7Exploration versus exploitation
  8. 1.8Formulating control problems as MDPs
  9. 1.9Taxonomy of reinforcement learning algorithms
  10. 1.10Case study: formulating an MDP for temperature control
  1. 2.1Value iteration and policy iteration
  2. 2.2Monte Carlo methods
  3. 2.3Temporal-difference learning TD(0)
  4. 2.4SARSA and Q-learning
  5. 2.5Eligibility traces
  6. 2.6Multi-armed bandits and UCB
  7. 2.7Convergence of tabular methods
  8. 2.8Reward design
  9. 2.9Simulation environments with Gymnasium
  10. 2.10Lab: Q-learning in a grid-world environment
  1. 3.1Function approximation with neural networks
  2. 3.2Deep Q-Network architecture
  3. 3.3Experience replay and target networks
  4. 3.4Double DQN and Dueling DQN
  5. 3.5Prioritised experience replay
  6. 3.6Rainbow and distributional extensions
  7. 3.7Training stability
  8. 3.8Critical training hyperparameters
  9. 3.9Evaluation and reproducibility
  10. 3.10Practical: DQN for a discrete control problem
  1. 4.1Policy gradient theorem
  2. 4.2REINFORCE and baselines
  3. 4.3A2C actor-critic methods
  4. 4.4Proximal Policy Optimisation (PPO)
  5. 4.5Trust region methods (TRPO)
  6. 4.6Continuous control with DDPG and TD3
  7. 4.7Soft Actor-Critic and entropy regularisation
  8. 4.8Parallelising training
  9. 4.9Reference implementations in Stable-Baselines3
  10. 4.10Lab: continuous control of a pendulum with SAC
  1. 5.1Learned dynamics models
  2. 5.2Planning with models
  3. 5.3Predictive control with learned models
  4. 5.4Dyna and world models
  5. 5.5MuZero and latent planning
  6. 5.6Model uncertainty and ensembles
  7. 5.7Sample efficiency of the methods
  8. 5.8System identification for RL
  9. 5.9Combining with physical models
  10. 5.10Practical: predictive control with a neural model
  1. 6.1Feedback, stability and robustness
  2. 6.2PID control and tuning
  3. 6.3State space and controllability
  4. 6.4Linear quadratic regulator
  5. 6.5Model predictive control
  6. 6.6Lyapunov stability
  7. 6.7H-infinity robust control
  8. 6.8Model reference adaptive control
  9. 6.9Comparing classical and learned control
  10. 6.10Workshop: LQR versus PPO on a mechanical system
  1. 7.1Defining safety in learned systems
  2. 7.2Constrained MDPs
  3. 7.3Constrained policy optimisation
  4. 7.4Safety shields
  5. 7.5Control barrier functions
  6. 7.6Safe exploration during learning
  7. 7.7Learned Lyapunov functions
  8. 7.8Misspecified rewards and reward hacking
  9. 7.9Safe RL benchmarks
  10. 7.10Lab: safety shield over a PPO controller
  1. 8.1From the simulator to the real world
  2. 8.2Domain randomisation
  3. 8.3Adversarial and robust RL
  4. 8.4Generalisation to new tasks
  5. 8.5Meta-learning for rapid adaptation
  6. 8.6Offline reinforcement learning
  7. 8.7Conservatism with static data
  8. 8.8Off-policy evaluation
  9. 8.9Policy stress testing
  10. 8.10Practical: sim-to-real transfer with randomisation
  1. 9.1Formal verification of neural networks
  2. 9.2Reachability analysis
  3. 9.3Temporal logic and specifications
  4. 9.4Scenario-based testing
  5. 9.5Controller falsification
  6. 9.6Runtime monitors
  7. 9.7Robustness certification
  8. 9.8Neural network verification tools
  9. 9.9Limits of verification
  10. 9.10Workshop: falsifying a controller with STL specifications
  1. 10.1Functional safety according to IEC 61508
  2. 10.2ISO 26262 and SOTIF ISO 21448
  3. 10.3ISO/IEC TR 5469 on AI and functional safety
  4. 10.4Safety life cycle with learned components
  5. 10.5Safety arguments and GSN
  6. 10.6Risk analysis of autonomous systems
  7. 10.7UL 4600 for autonomous systems
  8. 10.8Human oversight of autonomous systems
  9. 10.9AI Act and high-risk systems
  10. 10.10Case study: safety case for a learned controller
  1. 11.1Robotics and manipulation
  2. 11.2Autonomous vehicles and drones
  3. 11.3HVAC and building efficiency
  4. 11.4Power grids and storage
  5. 11.5Chemical process control
  6. 11.6Logistics and inventory management
  7. 11.7Sequential recommender systems
  8. 11.8Aligning language models with RLHF
  9. 11.9Lessons from real deployments
  10. 11.10Practical: HVAC control with constrained RL
  1. 12.1Defining the control problem
  2. 12.2Simulation environment
  3. 12.3Formulating rewards and constraints
  4. 12.4Policy training
  5. 12.5Safety mechanisms
  6. 12.6Verification and testing
  7. 12.7Evaluation against classical controllers
  8. 12.8Conclusions on safety and performance
  9. 12.9Master's Thesis report
  10. 12.10Master's Thesis: defence before a panel
Methodology, software & labs

How you'll learn

100% online training on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. There are no fixed timetables: you progress at your own pace within the intake, so you can combine the programme with your job.

Each module combines theory, guided reading of the original standard and applied exercises. The working software is Gymnasium, Stable-Baselines3, MATLAB Reinforcement Learning Toolbox, MuJoCo and Ray RLlib, in free, academic or evaluation versions. Real procedures are studied: continuous control of mechanical systems in simulation, energy management of buildings under constraints, and policy verification through falsification. Assessment is continuous by module, plus the final Master's Thesis.

Standards covered: IEC 61508, ISO 21448, ISO/IEC TR 5469, ISO 26262, Regulation (EU) 2024/1689 (AI Act), ANSI/UL 4600 and ISO/IEC 23894.

Assigned technical tutor:every module ends with a deliverable that your tutor marks individually.
Standards-based learning:guided reading of IEC 61508, ISO 21448 and ISO/IEC TR 5469 applied to real cases, not as a theoretical annex.
Industry software:Gymnasium, Stable-Baselines3, MATLAB Reinforcement Learning Toolbox, MuJoCo and Ray RLlib, in free, academic or evaluation versions.
Master's Thesis with defence:a safe learned controller, trained, verified and evaluated against classical controllers, defended before a panel.
Flexible format:100% online, no fixed timetables, compatible with full-time work.
Capstone project

Build it end to end

Your final project follows the real engineering workflow, step by step, reviewed by your tutor.

Define the control problem
Formulate the task as an MDP with rewards and safety constraints.
Train the policy
Train controllers with PPO or SAC in a simulation environment.
Add safety mechanisms
Integrate a safety shield or control barrier functions over the learned controller.
Verify and benchmark
Falsify the controller with STL specifications and compare it against LQR.
Master's Thesis
Present the safety case and performance conclusions, defended before a panel.
Career opportunities

Where this takes you

Reinforcement Learning EngineerControl EngineerSafe AI ResearcherAutonomous Systems EngineerFunctional Safety EngineerDecision Data Scientist
Admissions & scholarships

Requirements

  • University degree in engineering, science or architecture, or equivalent demonstrable professional experience.
  • Internet connection and a computer capable of running the programme software.
  • English sufficient to read technical standards.

Ready to start?

Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.

Apply now Scholarships
FAQ

Frequently asked questions

The programme totals 1,900 hours over 12–14 months. You set your own pace within each intake, and your tutor helps you plan the workload around your job.

You work with Gymnasium, Stable-Baselines3, MATLAB Reinforcement Learning Toolbox, MuJoCo and Ray RLlib, using free, academic or evaluation versions. Your tutor indicates the free alternative wherever one exists.

A SEIUM Institute diploma, awarded on passing the continuous assessment and the Master's Thesis. It is SEIUM's own qualification, not an official state degree; its recognition depends on the programme completed and the regulations of each country.

Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the higher programme. Ask about the recognition policy when you enrol.

A university degree in engineering, science or architecture (or equivalent professional experience), a computer able to run the software, and English sufficient to read technical standards.

Start dates are set by SEIUM's faculty for each intake. Our admissions team will confirm the next available date when you apply.

Keep exploring

Related programmes

See all in Data, AI, Simulation & Analytics
Master's9,970 $

Master’s in Advanced Vision/Geomatics Engineering (Beyond Mobility)

Online master's programme of 1,900 hours in advanced vision and geomatics engineering beyond mobility. 12 modules covering reference systems and geodesy, optical remote sensing, SAR and…

1900 hOnlineData, AI, Simulation & Analytics
Master's9,970 $

Master’s in Optimisation, Operations Research and Decision Engineering

Online master's programme of 1,900 hours in optimisation, operations research and decision-making: 12 modules covering modelling, linear, integer and non-linear programming, metaheuristics, constraint programming and scheduling,…

1900 hOnlineData, AI, Simulation & Analytics
Master's11,090 $

Master’s in Graph ML Engineering for Networks and Systems

Online master's programme of 1,900 hours in graph machine learning engineering for networks and systems. 12 modules covering engineering systems as graphs, classical network analysis, embeddings,…

1900 hOnlineData, AI, Simulation & Analytics
Master’s in Reinforcement Learning, Control and Safe AI EngineeringEnrol
Related programmes

More in Data, AI, Simulation & Analytics

Master’s in Time Series, Signals and Anomaly Detection EngineeringView programme →Master’s in AI Governance and Compliance EngineeringView programme →Master’s in Multiphysics Co-Simulation & FMI EngineeringView programme →Master’s in Synthetic Data and Sensor Simulation EngineeringView programme →Master’s in Model Verification, Validation and UQ EngineeringView programme →Master’s in Optimisation, Operations Research and Decision EngineeringView programme →

All Data, AI, Simulation & Analytics programmes → · How to apply · Fees & financing · Scholarships

Scroll to Top
SEIUM – Institute of Advanced Engineering
Powered by  GDPR Cookie Compliance
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.