NeurIPS 2023: Safety-Gymnasium: A Unified Safe Reinforcement Learning Benchmark

A framework for the elicitation, specification, formalization and analysis of requirements.

Constant-complexity, deterministic, very fast memory allocator (heap) for hard real-time high-integrity embedded systems. Allocation takes ≈165 cycles on Cortex M4. There is little activity because the project is finished and does not require further changes.

A collection of work using nonlinear model predictive control (NMPC) with discrete-time control Lyapunov functions (CLFs) and control barrier functions (CBFs)

"Safety-Critical Model Predictive Control with Discrete-Time Control Barrier Function" by J. Zeng, B. Zhang and K. Sreenath https://arxiv.org/abs/2007.11718

Safe robot learning

ICLR 2024: SafeDreamer: Safe Reinforcement Learning with World Models

Safe Pontryagin Differentiable Programming (Safe PDP) is a new theoretical and algorithmic safe differentiable framework to solve a broad class of safety-critical learning and control tasks.

The Fuzion Language Implementation

"Safety-Critical Control using Optimal-decay Control Barrier Functions with Guaranteed Point-wise Feasibility" by J. Zeng, B. Zhang, Z. Li and K. Sreenath https://arxiv.org/pdf/2103.12375.pdf

Code for the paper "Control Barriers in Bayesian Learning of System Dynamics"

Real Time Safety Heap Allocator

List of NSO Pegasus's C2C Servers where will be used to control Pegasus.

Code for L4DC 2022 paper: Joint Synthesis of Safety Certificate and Safe Control Policy Using Constrained Reinforcement Learning.

Safety Critical Control of Autonomous Vehicles by Control Barrier Functions

🚧 🚔 ⚠ Toolbox to compute Criticality Measures for Automated Vehicles

This extended Eigen C++ template library and wrapper provide a malloc-free Moore-Penrose pseudoinverse solver.

On the forward invariance of Neural ODEs: performance guarantees for policy learning

🩺🛣️ IBP IoU an approach for the formal verificaion of object detection models.

Implementation of the Heatmap-based Unsupervised Debugging of DNNs (HUDD) toolset