Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Operational Safety in Human-in-the-loop Human-in-the-plant Autonomous Systems

Ayan Banerjee, Aranyak Maity, Imane Lamrani, Sandeep K. S. Gupta

TL;DR

It is shown that synthesized HIL-HIP controller for automated insulin delivery in Type 1 Diabetes is the only controller to meet safety requirements for human action inputs.

Abstract

Control affine assumptions, human inputs are external disturbances, in certified safe controller synthesis approaches are frequently violated in operational deployment under causal human actions. This paper takes a human-in-the-loop human-in-the-plant (HIL-HIP) approach towards ensuring operational safety of safety critical autonomous systems: human and real world controller (RWC) are modeled as a unified system. A three-way interaction is considered: a) through personalized inputs and biological feedback processes between HIP and HIL, b) through sensors and actuators between RWC and HIP, and c) through personalized configuration changes and data feedback between HIL and RWC. We extend control Lyapunov theory by generating barrier function (CLBF) under human action plans, model the HIL as a combination of Markov Chain for spontaneous events and Fuzzy inference system for event responses, the RWC as a black box, and integrate the HIL-HIP model with neural architectures that can learn CLBF certificates. We show that synthesized HIL-HIP controller for automated insulin delivery in Type 1 Diabetes is the only controller to meet safety requirements for human action inputs.

Operational Safety in Human-in-the-loop Human-in-the-plant Autonomous Systems

TL;DR

It is shown that synthesized HIL-HIP controller for automated insulin delivery in Type 1 Diabetes is the only controller to meet safety requirements for human action inputs.

Abstract

Control affine assumptions, human inputs are external disturbances, in certified safe controller synthesis approaches are frequently violated in operational deployment under causal human actions. This paper takes a human-in-the-loop human-in-the-plant (HIL-HIP) approach towards ensuring operational safety of safety critical autonomous systems: human and real world controller (RWC) are modeled as a unified system. A three-way interaction is considered: a) through personalized inputs and biological feedback processes between HIP and HIL, b) through sensors and actuators between RWC and HIP, and c) through personalized configuration changes and data feedback between HIL and RWC. We extend control Lyapunov theory by generating barrier function (CLBF) under human action plans, model the HIL as a combination of Markov Chain for spontaneous events and Fuzzy inference system for event responses, the RWC as a black box, and integrate the HIL-HIP model with neural architectures that can learn CLBF certificates. We show that synthesized HIL-HIP controller for automated insulin delivery in Type 1 Diabetes is the only controller to meet safety requirements for human action inputs.
Paper Structure (17 sections, 3 theorems, 5 equations, 7 figures)

This paper contains 17 sections, 3 theorems, 5 equations, 7 figures.

Table of Contents

  1. Introduction
  2. System Model and preliminaries
  3. Formal Problem Statement
  4. Solution and proof of safety
  5. Modeling external events with Markov Chain
  6. Finding DoA of MC with minimum probability $p$
  7. Learning a FIS from data
  8. Finding reach set for FIS
  9. Safety certificate generation method
  10. Evaluation and Results
  11. Data Description
  12. Safety violation in control affine assumption
  13. FIS model accuracy
  14. MC modeling accuracy
  15. Performance of safety certified HIL-HIP system
  16. ...and 2 more sections

Key Result

Lemma 1

The set of states explored by the optimal policy with value function $v^*(e) > p$, gives the reach set of the MC $(S,P_S)$ starting from state $e$, where $v^*$ is given by Eqn eqn:MDPV.

Figures (7)

  • Figure 1: HIL-HIP autonomous systems (AS).
  • Figure 2: Categorization of HIL actions.
  • Figure 3: Solution approach
  • Figure 4: Solution method for deriving safety certificate for HIL=HIP systems under the learned human aciton model
  • Figure 5: Safety violations occur if human inputs are considered as external disturbances in AID systems for T1D.
  • ...and 2 more figures

Theorems & Definitions (3)

  • Lemma 1
  • Lemma 2
  • Theorem 1