Safe learning in robotics: From learning-based control to safe reinforcement learning

L Brunke, M Greeff, AW Hall, Z Yuan… - Annual Review of …, 2022 - annualreviews.org
The last half decade has seen a steep rise in the number of contributions on safe learning
methods for real-world robotic deployments from both the control and reinforcement learning …

On the opportunities and risks of foundation models

R Bommasani, DA Hudson, E Adeli, R Altman… - arXiv preprint arXiv …, 2021 - arxiv.org
AI is undergoing a paradigm shift with the rise of models (eg, BERT, DALL-E, GPT-3) that are
trained on broad data at scale and are adaptable to a wide range of downstream tasks. We …

Robust control barrier–value functions for safety-critical control

JJ Choi, D Lee, K Sreenath, CJ Tomlin… - 2021 60th IEEE …, 2021 - ieeexplore.ieee.org
This paper works towards unifying two popular approaches in the safety control community:
Hamilton-Jacobi (HJ) reachability and Control Barrier Functions (CBFs). HJ Reachability has …

Data-driven safety filters: Hamilton-jacobi reachability, control barrier functions, and predictive methods for uncertain systems

KP Wabersich, AJ Taylor, JJ Choi… - IEEE Control …, 2023 - ieeexplore.ieee.org
Today's control engineering problems exhibit an unprecedented complexity, with examples
including the reliable integration of renewable energy sources into power grids, safe …

Refining control barrier functions through Hamilton-Jacobi reachability

S Tonkens, S Herbert - 2022 IEEE/RSJ International …, 2022 - ieeexplore.ieee.org
Safety filters based on Control Barrier Functions (CBFs) have emerged as a practical tool for
the safety-critical control of autonomous systems. These approaches encode safety through …

Synthesizing control barrier functions with feasible region iteration for safe reinforcement learning

Y Yang, Y Zhang, W Zou, J Chen… - IEEE Transactions on …, 2023 - ieeexplore.ieee.org
Safety is a critical concern when applying reinforcement learning to real-world control
problems. A widely used method for ensuring safety is to learn a control barrier function with …

The Feasibility of Constrained Reinforcement Learning Algorithms: A Tutorial Study

Y Yang, Z Zheng, SE Li, M Tomizuka, C Liu - arXiv preprint arXiv …, 2024 - arxiv.org
Satisfying safety constraints is a priority concern when solving optimal control problems
(OCPs). Due to the existence of infeasibility phenomenon, where a constraint-satisfying …

Provably safe reinforcement learning: A theoretical and experimental comparison

H Krasowski, J Thumm, M Müller, L Schäfer… - arXiv preprint arXiv …, 2022 - arxiv.org
Ensuring safety of reinforcement learning (RL) algorithms is crucial to unlock their potential
for many real-world tasks. However, vanilla RL does not guarantee safety. In recent years …

Constructing control lyapunov-value functions using hamilton-jacobi reachability analysis

Z Gong, M Zhao, T Bewley… - IEEE Control Systems …, 2022 - ieeexplore.ieee.org
In this letter, we seek to build connections between control Lyapunov functions (CLFs) and
Hamilton-Jacobi (HJ) reachability analysis. CLFs have been used extensively in the control …

Safe non-stochastic control of control-affine systems: An online convex optimization approach

H Zhou, Y Song, V Tzoumas - IEEE Robotics and Automation …, 2023 - ieeexplore.ieee.org
We study how to safely control nonlinear control-affine systems that are corrupted with
bounded non-stochastic noise, ie, noise that is unknown a priori and that is not necessarily …