Lecture

Safe Learning in Robotics: Quo Vadis?

Graph Chat

Chat with Graph Search

Ask any question about EPFL courses, lectures, exercises, research, news, etc. or try the example questions below.

DISCLAIMER: The Graph Chatbot is not programmed to provide explicit or categorical answers to your questions. Rather, it transforms your questions into API requests that are distributed across the various IT services officially administered by EPFL. Its purpose is solely to collect and recommend relevant references to content that you can explore to help you answer your questions.

Description

This lecture by the instructor covers the topic of safe learning in robotics, focusing on the state of the art, enabling progress, open challenges, and vision in the field. The lecture discusses the safe learning control problem, types of safety guarantees, and approaches to encouraging safety and robustness in reinforcement learning. It also explores certifying learning-based control and constraint set certification. The presentation includes key features of safe-control-gym, current controllers, results on constraint satisfaction, performance, and data efficiency, as well as the IROS 2022 Safe Robot Learning Competition. The lecture concludes with a summary of the challenges in the field and the importance of interdisciplinary collaboration for safe robot autonomy research.

Official source

Related concepts (32)

Reinforcement learning

Reinforcement learning (RL) is an area of machine learning concerned with how intelligent agents ought to take actions in an environment in order to maximize the notion of cumulative reward. Reinforcement learning is one of three basic machine learning paradigms, alongside supervised learning and unsupervised learning. Reinforcement learning differs from supervised learning in not needing labelled input/output pairs to be presented, and in not needing sub-optimal actions to be explicitly corrected.

Control system

A control system manages, commands, directs, or regulates the behavior of other devices or systems using control loops. It can range from a single home heating controller using a thermostat controlling a domestic boiler to large industrial control systems which are used for controlling processes or machines. The control systems are designed via control engineering process. For continuously modulated control, a feedback controller is used to automatically control a process or operation.

Control engineering

Control engineering or control systems engineering is an engineering discipline that deals with control systems, applying control theory to design equipment and systems with desired behaviors in control environments. The discipline of controls overlaps and is usually taught along with electrical engineering and mechanical engineering at many institutions around the world. The practice uses sensors and detectors to measure the output performance of the process being controlled; these measurements are used to provide corrective feedback helping to achieve the desired performance.

Control theory

Control theory is a field of control engineering and applied mathematics that deals with the control of dynamical systems in engineered processes and machines. The objective is to develop a model or algorithm governing the application of system inputs to drive the system to a desired state, while minimizing any delay, overshoot, or steady-state error and ensuring a level of control stability; often with the aim to achieve a degree of optimality. To do this, a controller with the requisite corrective behavior is required.

Optimal control

Optimal control theory is a branch of mathematical optimization that deals with finding a control for a dynamical system over a period of time such that an objective function is optimized. It has numerous applications in science, engineering and operations research. For example, the dynamical system might be a spacecraft with controls corresponding to rocket thrusters, and the objective might be to reach the moon with minimum fuel expenditure.

Related lectures (64)

Acquiring Data for Learning: Modern Approaches and Challenges

Explores modern approaches and challenges in acquiring data for learning optimal controllers through demonstrations and data-driven methods.

Acquiring Data for Learning

Explores training robots through reinforcement learning and learning from demonstration, highlighting challenges in human-robot interaction and data collection.

Perceptual Robotics: Integrating Vision and Action

Covers the integration of visual perception and robotic actions in embodied AI.

Safe Learning in Automatic Control Systems

Explores safe learning in automatic control systems, covering challenges, risk-sensitive control, and predictive safety filters.

Force Control with Dynamical Systems