Dynamic-weighted simplex strategy for learning enabled cyber physical systems

Shreyas Ramakrishna, Charles Harstell, Matthew P. Burruss, Gabor Karsai, Abhishek Dubey

Journal of Systems Architecture 2020

PDF DOI

Why This Matters

Learning-enabled components offer impressive capabilities but present challenges in ensuring system safety and correctness under all circumstances. This work is innovative because it combines formal safety assurance from the Simplex Architecture with machine learning optimization, enabling systems to leverage high-performance learned components while maintaining provable safety guarantees. The dynamic weighting mechanism allows gradual transitions that avoid abrupt control switches.

What We Did

This paper introduces a dynamic-weighted simplex strategy for learning-enabled cyber-physical systems that use Learning Enabled Components for autonomous control. The approach extends the classical Simplex Architecture by incorporating reinforcement learning to dynamically weight controller outputs, allowing smooth transitions between an advanced high-performance controller and a safe baseline controller. The framework was demonstrated on a DeepNNCar autonomous vehicle platform with real-time performance monitoring and resource management.

Key Results

The dynamic-weighted simplex strategy achieved 60% fewer out-of-track soft constraint violations compared to the original LEC-driven system while demonstrating higher optimized speed performance of 0.4 m/s during indoor driving. The approach successfully reduced computational overhead by avoiding the full complexity of the advanced controller when not needed. The framework proved effective at balancing safety and performance in real-world autonomous driving scenarios.

Full Abstract

Cite This Paper

@article{ramakrishna2020dynamic,
  author = {Ramakrishna, Shreyas and Harstell, Charles and Burruss, Matthew P. and Karsai, Gabor and Dubey, Abhishek},
  journal = {Journal of Systems Architecture},
  title = {Dynamic-weighted simplex strategy for learning enabled cyber physical systems},
  year = {2020},
  issn = {1383-7621},
  pages = {101760},
  volume = {111},
  abstract = {Cyber Physical Systems (CPS) have increasingly started using Learning Enabled Components (LECs) for performing perception-based control tasks. The simple design approach, and their capability to continuously learn has led to their widespread use in different autonomous applications. Despite their simplicity and impressive capabilities, these components are difficult to assure, which makes their use challenging. The problem of assuring CPS with untrusted controllers has been achieved using the Simplex Architecture. This architecture integrates the system to be assured with a safe controller and provides a decision logic to switch between the decisions of these controllers. However, the key challenges in using the Simplex Architecture are: (1) designing an effective decision logic, and (2) sudden transitions between controller decisions lead to inconsistent system performance. To address these research challenges, we make three key contributions: (1) dynamic-weighted simplex strategy – we introduce “weighted simplex strategy” as the weighted ensemble extension of the classical Simplex Architecture. We then provide a reinforcement learning based mechanism to find dynamic ensemble weights, (2) middleware framework – we design a framework that allows the use of the dynamic-weighted simplex strategy, and provides a resource manager to monitor the computational resources, and (3) hardware testbed – we design a remote-controlled car testbed called DeepNNCar to test and demonstrate the aforementioned key concepts. Using the hardware, we show that the dynamic-weighted simplex strategy has 60% fewer out-of-track occurrences (soft constraint violations), while demonstrating higher optimized speed (performance) of 0.4 m/s during indoor driving than the original LEC driven system.},
  contribution = {lead},
  doi = {https://doi.org/10.1016/j.sysarc.2020.101760},
  keywords = {learning-enabled components, simplex architecture, autonomous vehicles, reinforcement learning, safety assurance, control synthesis},
  tag = {a14cps},
  url = {https://www.sciencedirect.com/science/article/pii/S1383762120300540}
}
Quick Info
Year 2020
DOI https://doi.org/10.1016/j.s...
Keywords
learning-enabled components simplex architecture autonomous vehicles reinforcement learning safety assurance control synthesis
Research Areas
CPS ML for CPS Explainable AI
Search Tags

Dynamic, weighted, simplex, strategy, learning, enabled, cyber, physical, systems, learning-enabled components, simplex architecture, autonomous vehicles, reinforcement learning, safety assurance, control synthesis, CPS, ML for CPS, Explainable AI, 2020, Ramakrishna, Harstell, Burruss, Karsai, Dubey