Safe CPS from unsafe controllers

  title={Safe CPS from unsafe controllers},
  author={Usama Mehmood and Stanley Bak and Scott A. Smolka and Scott D. Stoller},
  journal={Proceedings of the Workshop on Computation-Aware Algorithmic Design for Cyber-Physical Systems},
  • Usama Mehmood, Stanley Bak, S. Stoller
  • Published 24 February 2021
  • Computer Science
  • Proceedings of the Workshop on Computation-Aware Algorithmic Design for Cyber-Physical Systems
Modern cyber-physical systems (CPS) interact with the physical world, hence their correctness is important. In this work, we build upon the Simplex Architecture, where control authority may switch from an unverified and potentially unsafe advanced controller to a verified-safe baseline controller in order to maintain system safety. We take the approach further by lifting the requirement that the baseline controller must be verified or even correct, instead also treating it as a black-box… 

Figures from this paper

Run Time Assurance for Safety-Critical Systems: An Introduction to Safety Filtering Approaches for Complex Control Systems
The Automatic Ground Collision Avoidance System (Auto GCAS), an RTA system integrated on the jets less than two years earlier in the Fall of 2014, detected that the aircraft was about to collide, commanded a roll to wings level and pull up maneuver, and recovered the aircraft less than 3,000 feet above the Arizona desert.
Runtime Safety Assurance for Learning-enabled Control of Autonomous Driving Vehicles
Simulation experiment results demonstrate that Simplex-Drive can always ensure the operation safety without sacrificing control performance, even if the DRL policy may lead to deviations from the safe status.


Safe CPS from Unsafe Controllers
  • 2021
ARCH-COMP20 Category Report: Continuous and Hybrid Systems with Nonlinear Dynamics
The results of a friendly competition for formal verification of continuous and hybrid systems with nonlinear continuous dynamics with Ariadne, CORA, DynIbex, Flow*, Isabelle/HOL, and JuliaReach participated and the potential advantages of different tools are discovered.
Improved Geometric Path Enumeration for Verifying ReLU Neural Networks
This paper works to address the runtime problem by improving upon a recently-proposed geometric path enumeration method, and demonstrates significant speed improvement of exact analysis on the well-studied ACAS Xu benchmarks, sometimes hundreds of times faster than the original implementation.
Neural Simplex Architecture
The Neural Simplex Architecture (NSA), a new approach to runtime assurance that provides safety guarantees for neural controllers of autonomous and other complex systems without unduly sacrificing performance, is presented.
ReachFlow: An Online Safety Assurance Framework for Waypoint-Following of Self-driving Cars
This work proposes an online monitor called Reach-Flow for fault prevention of waypoint-following tasks for self-driving cars and demonstrates the effectiveness by rigorously verifying a safe way point- following control and providing a fallback control for an unsafe situation in which a large deviation from the planned path is predicted.
Contingency Model Predictive Control for Automated Vehicles
We present Contingency Model Predictive Control (CMPC), a novel and implementable control framework which tracks a desired path while simultaneously maintaining a contingency plan - an alternate
Deep Neural Network Compression for Aircraft Collision Avoidance Systems
A deep neural network is used to approximate the table, reducing the required storage space by a factor of 1000 and enabling the collision avoidance system to operate using current avionics systems.
SOTER: A Runtime Assurance Framework for Programming Safe Robotics Systems
SOTER is presented, a robotics programming framework with two key components: a programming language for implementing and testing high-level reactive robotics software, and an integrated runtime assurance (RTA) system that helps enable the use of uncertified components, while still providing safety guarantees.
An Online Approach to Active Set Invariance
This paper shows how one can constrain a system to stay within reach of an appropriately chosen backup set in a minimally invasive way by performing online sensitivity analysis around a backup trajectory and factor in state constraints to enforce set invariance only based on online computations of sensitivities.
DeepPicar: A Low-Cost Deep Neural Network-Based Autonomous Car
  • M. Bechtel, Elise McEllhiney, H. Yun
  • Computer Science
    2018 IEEE 24th International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA)
  • 2018
Using DeepPicar, a low-cost deep neural network based autonomous car platform, the Pi 3's computing capabilities to support end-to-end deep learning based real-time control of autonomous vehicles are analyzed and state-of-the-art cache partitioning and memory bandwidth throttling techniques are evaluated.