Master’s in Reinforcement & Imitation Learning for Control Engineering
Online master's programme of 1,900 hours in reinforcement and imitation learning for control. 12 modules covering RL algorithms for robots, imitation and inverse reinforcement learning, demonstration capture, sim-to-real transfer, visuomotor policies and safe low-level control under ISO 10218-1, ISO/TS 15066 and Regulation (EU) 2023/1230, with an assigned technical tutor.
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.
What you will be able to do
Formulate robotic tasks as learning problems.
Train control policies through reinforcement learning.
Apply imitation techniques based on human demonstrations.
Infer reward functions through inverse learning.
Transfer policies from simulation to the real robot.
Integrate learned policies with safe low-level controllers.
Assess the risks and regulatory compliance of learning robots.
Who it's for
- Robot learning engineers who want to master both imitation and reinforcement approaches.
- Control engineers looking to combine classical controllers with learned policies.
- Robotics researchers seeking an industrial, standards-aware perspective.
- Simulation engineers working on sim-to-real pipelines.
- Collaborative robot integrators who need to deploy learning-based behaviours safely.
- Vision engineers for robotics moving into visuomotor control.
- Engineering or science graduates looking to enter industrial robotics.
What you'll work with
- ISO 10218-1 (safety requirements for industrial robots)
- ISO/TS 15066 (collaborative robots)
- Regulation (EU) 2023/1230 (Machinery Regulation)
- ISO 10218-2 (robot systems and integration)
- Regulation (EU) 2024/1689 (AI Act)
- ISO 13849-1 (safety-related parts of control systems)
- ISO 12100 (risk assessment and risk reduction of machinery)
- MuJoCo
- NVIDIA Isaac Sim
- ROS 2
- LeRobot
- Stable-Baselines3
12 modules · 1900 h
1.1From programmed to learned controllers1.2State, action and observation spaces in robots1.3Dynamics of manipulators and vehicles1.4Perception for control1.5Manipulation, locomotion and navigation tasks1.6Human demonstrations as a data source1.7Physics simulators for robotics1.8Success metrics in robotic tasks1.9State of the art in robot learning1.10Case study: analysis of a robotic assembly task
2.1MDPs and value functions2.2Q-learning and temporal differences2.3Policy gradient2.4PPO for continuous control2.5SAC and off-policy learning2.6Designing dense and sparse rewards2.7Training curricula2.8Massive parallelisation on GPU2.9Diagnosing failed training runs2.10Lab: quadruped locomotion with PPO
3.1Behaviour cloning3.2Covariate shift3.3DAgger and interactive correction3.4Multi-modal imitation3.5Diffusion policies3.6Action transformers3.7Imitation from observations3.8Quality and diversity of demonstrations3.9Evaluating imitated policies3.10Practical: behaviour cloning for pick and place
4.1Inferring the reward from the expert4.2Maximum entropy IRL4.3Generative adversarial imitation learning (GAIL)4.4AIRL and transferable rewards4.5Preference learning4.6Rewards from video4.7Reward ambiguity4.8Applications in driving4.9Risks of learned rewards4.10Lab: GAIL in a continuous control environment
5.1Teleoperation with haptic devices5.2Kinaesthetic teaching by guiding the arm5.3Virtual reality for teleoperation5.4Human motion capture5.5Human-to-robot retargeting5.6Low-cost collection systems5.7Synchronising sensors and actions5.8Annotation and segmentation of demonstrations5.9Open robotics datasets5.10Workshop: teleoperation system for collecting demonstrations
6.1Pre-training by imitation and fine-tuning by reinforcement6.2Demonstrations in the replay buffer6.3Offline RL with expert data6.4Regularisation towards the expert policy6.5Residual learning on top of classical controllers6.6Human correction during deployment6.7Reinforcement learning from human feedback6.8Efficiency with real-world samples6.9Comparison of approaches6.10Practical: residual policy on top of an impedance controller
7.1MuJoCo, Isaac Sim and PyBullet simulators7.2Contact and friction modelling7.3Physical and visual domain randomisation7.4System identification to close the gap7.5Online adaptation7.6Digital twins of cells7.7Large-scale GPU simulation7.8Validating the transfer7.9Typical errors in sim-to-real7.10Lab: transferring a grasping policy
8.1Policies from pixels8.2Pre-trained visual representations8.3Fusion of vision, force and proprioception8.4Point clouds for manipulation8.5Self-supervised representation learning8.6Vision-language-action models8.7Visual robustness to lighting and occlusions8.8Latency and control frequency8.9Camera and robot calibration8.10Workshop: visuomotor policy for pushing objects
9.1Impedance and admittance control9.2Force control in contact9.3Interface between learned policy and controller9.4Speed, force and workspace limits9.5Collaborative robots according to ISO/TS 150669.6Requirements of ISO 102189.7Safety filters and barrier functions9.8Runtime monitoring9.9Safe stop and recovery9.10Practical: safety filter for a collaborative arm
10.1Assembly with precision insertion10.2Manipulation of deformable objects10.3Bin picking10.4Welding and contact processes10.5Mobile robot navigation10.6Autonomous driving by imitation10.7Drones and agile control10.8Exoskeletons and assistive robotics10.9Economic viability of robot learning10.10Case study: learned peg-in-hole insertion
11.1Reproducible evaluation protocols11.2Generalisation to new objects and environments11.3Testing on real hardware11.4Maintaining policies in production11.5Machinery Regulation and learning systems11.6AI Act and robotics11.7Accountability and human oversight11.8Labour impact and acceptance11.9Technical documentation of the robotic system11.10Case study: risk assessment of a learning robot
12.1Defining the robotic task12.2Simulation environment or real test bench12.3Collecting demonstrations12.4Imitation training12.5Reinforcement refinement12.6Transfer and testing12.7Safety and performance evaluation12.8Conclusions and future work12.9Master's Thesis report12.10Master's Thesis: defence before a panel
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 MuJoCo, NVIDIA Isaac Sim, ROS 2, LeRobot and Stable-Baselines3, in free, academic or evaluation versions. Real procedures are studied: behaviour cloning in manipulation tasks, transferring policies from simulation to a real robot, and teleoperation for collecting demonstrations. Assessment is continuous by module, plus the final Master's Thesis.
Standards covered: ISO 10218-1, ISO/TS 15066, Regulation (EU) 2023/1230 (Machinery), ISO 10218-2, Regulation (EU) 2024/1689 (AI Act), ISO 13849-1 and ISO 12100.
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Where this takes you
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.
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 MuJoCo, NVIDIA Isaac Sim, ROS 2, LeRobot and Stable-Baselines3, 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 programme 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.
Related programmes
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,…
Master’s in Data, Real-Time and Platforms Engineering
Online master's programme of 1,900 hours in data engineering, real-time processing and platforms: 12 modules covering modern data architectures, batch and streaming processing, orchestration, data quality…
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…
More in Data, AI, Simulation & Analytics
All Data, AI, Simulation & Analytics programmes → · How to apply · Fees & financing · Scholarships