Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Closed-loop multi-step planning with innate physics knowledge

Giulia Lafratta, Bernd Porr, Christopher Chandler, Alice Miller

TL;DR

This work presents a hierarchical framework to solve robot planning as an input control problem, and implements this framework on a real robot and test it in an overtaking scenario as proof-of-concept.

Abstract

We present a hierarchical framework to solve robot planning as an input control problem. At the lowest level are temporary closed control loops, ("tasks"), each representing a behaviour, contingent on a specific sensory input and therefore temporary. At the highest level, a supervising "Configurator" directs task creation and termination. Here resides "core" knowledge as a physics engine, where sequences of tasks can be simulated. The Configurator encodes and interprets simulation results,based on which it can choose a sequence of tasks as a plan. We implement this framework on a real robot and test it in an overtaking scenario as proof-of-concept.

Closed-loop multi-step planning with innate physics knowledge

TL;DR

This work presents a hierarchical framework to solve robot planning as an input control problem, and implements this framework on a real robot and test it in an overtaking scenario as proof-of-concept.

Abstract

We present a hierarchical framework to solve robot planning as an input control problem. At the lowest level are temporary closed control loops, ("tasks"), each representing a behaviour, contingent on a specific sensory input and therefore temporary. At the highest level, a supervising "Configurator" directs task creation and termination. Here resides "core" knowledge as a physics engine, where sequences of tasks can be simulated. The Configurator encodes and interprets simulation results,based on which it can choose a sequence of tasks as a plan. We implement this framework on a real robot and test it in an overtaking scenario as proof-of-concept.

Paper Structure

This paper contains 13 sections, 3 equations, 3 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Methods
  3. Preliminaries
  4. Tasks as closed-loop controllers
  5. The Configurator
  6. Formalism
  7. Implementation
  8. Core knowledge
  9. Reset
  10. Construction of cognitive map $\mathcal{G}$
  11. Guard $\Psi$
  12. Results
  13. Conclusion

Figures (3)

  • Figure 1: A: closed-loop obstacle avoidance. B: a closed-loop controller.
  • Figure 2: Example of the multi-step-ahead planning procedure carried out by the Configurator. A: a scenario requiring multi-step planning, B: Configurator, C: the plan, D: expansion of the start state, E: expansion of bext next state, F: full state-space expansion, G: state diagram of panel E, H: full state diagram.
  • Figure 3: Initial frame and tracking of the behaviour exhibited by the robot. Targets are depicted as hearts, collisions as stars. A: multiple-loop planning, B: one-loop-ahead disturbance detection, C: simulation trace and state labels for planning condition, executed plan highlighted in bold.

Theorems & Definitions (4)

  • Definition 1
  • Definition 2
  • Definition 3
  • Definition 4