Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Reference-Steering via Data-Driven Predictive Control for Hyper-Accurate Robotic Flying-Hopping Locomotion

Yicheng Zeng, Yuhao Huang, Xiaobin Xiong

TL;DR

The paper tackles the gap between model-based locomotion controllers and real hardware by introducing a data-driven input-output (DD-IO) reference-steering layer that leverages Data-Driven Predictive Control (DDPC) to online adjust reference trajectories so the realized output matches the target. By preserving the underlying model-based controller and augmenting it with a data-driven predictor, the approach enables hyper-accurate trajectory tracking for both flying and periodic hopping on the PogoX platform, including an artificial IO data generation mechanism to handle hybrid dynamics. The method is validated in simulation and hardware, showing significant improvements in tracking accuracy and disturbance rejection, and is presented as a general, plug-in augmentation for complex robotic systems. The work highlights a practical, data-driven paradigm for enhancing performance without re-synthesizing full model-based controllers, with potential extensions to nonlinear dynamics and terrain-adaptive operation.

Abstract

State-of-the-art model-based control designs have been shown to be successful in realizing dynamic locomotion behaviors for robotic systems. The precision of the realized behaviors in terms of locomotion performance via fly, hopping, or walking has not yet been well investigated, despite the fact that the difference between the robot model and physical hardware is doomed to produce inaccurate trajectory tracking. To address this inaccuracy, we propose a referencing-steering method to bridge the model-to-real gap by establishing a data-driven input-output (DD-IO) model on top of the existing model-based design. The DD-IO model takes the reference tracking trajectories as the input and the realized tracking trajectory as the output. By utilizing data-driven predictive control, we steer the reference input trajectories online so that the realized output ones match the actual desired ones. We demonstrate our method on the robot PogoX to realize hyper-accurate hopping and flying behaviors in both simulation and hardware. This data-driven reference-steering approach is straightforward to apply to general robotic systems for performance improvement via hyper-accurate trajectory tracking.

Reference-Steering via Data-Driven Predictive Control for Hyper-Accurate Robotic Flying-Hopping Locomotion

TL;DR

The paper tackles the gap between model-based locomotion controllers and real hardware by introducing a data-driven input-output (DD-IO) reference-steering layer that leverages Data-Driven Predictive Control (DDPC) to online adjust reference trajectories so the realized output matches the target. By preserving the underlying model-based controller and augmenting it with a data-driven predictor, the approach enables hyper-accurate trajectory tracking for both flying and periodic hopping on the PogoX platform, including an artificial IO data generation mechanism to handle hybrid dynamics. The method is validated in simulation and hardware, showing significant improvements in tracking accuracy and disturbance rejection, and is presented as a general, plug-in augmentation for complex robotic systems. The work highlights a practical, data-driven paradigm for enhancing performance without re-synthesizing full model-based controllers, with potential extensions to nonlinear dynamics and terrain-adaptive operation.

Abstract

State-of-the-art model-based control designs have been shown to be successful in realizing dynamic locomotion behaviors for robotic systems. The precision of the realized behaviors in terms of locomotion performance via fly, hopping, or walking has not yet been well investigated, despite the fact that the difference between the robot model and physical hardware is doomed to produce inaccurate trajectory tracking. To address this inaccuracy, we propose a referencing-steering method to bridge the model-to-real gap by establishing a data-driven input-output (DD-IO) model on top of the existing model-based design. The DD-IO model takes the reference tracking trajectories as the input and the realized tracking trajectory as the output. By utilizing data-driven predictive control, we steer the reference input trajectories online so that the realized output ones match the actual desired ones. We demonstrate our method on the robot PogoX to realize hyper-accurate hopping and flying behaviors in both simulation and hardware. This data-driven reference-steering approach is straightforward to apply to general robotic systems for performance improvement via hyper-accurate trajectory tracking.

Paper Structure

This paper contains 16 sections, 2 theorems, 10 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries
  4. Behavioral System Theory
  5. Data-Driven Predictive Control
  6. Flying-Hopping Control of PogoX
  7. Reference-Steering for Flying
  8. Flying Dynamics and Controller
  9. Reference-Steering via DDPC
  10. Reference-Steering for Periodic Hopping
  11. Hybrid Dynamics and Hopping Control
  12. Reference-Steering via DDPC
  13. Evaluation
  14. Reference-steering for Flying
  15. Reference-steering for Periodic Hopping
  16. ...and 1 more sections

Key Result

Theorem III.1

For a discrete-time LTI system described in eq:linearDyn, assume $(A,B)$ is controllable. Let $(u_{[0,T-1]},y_{[0,T-1]})$ be a sequence of input-output trajectory with $T, L \in \mathbb{Z}_{\geq 0}$. If $u_{[0,T-1]}$ is persistently exciting of order $L+n$, then for a sequence of new input-output tr

Figures (5)

  • Figure 1: Overview of the data-driven reference-steering approach (bottom) and the application on PogoX (top).
  • Figure 2: The control architecture of PogoX.
  • Figure 3: Illustration of artificial ground phase trajectory generation, where $u_{<d,a>\text{ini}}$ and $y_{<d,a>\text{ini}}$ represent the collected input and output data, and $u_{g-\text{ini}}^{pred},y_{g-\text{ini}}^{pred}$ represent the generated IO data of the ground phase.
  • Figure 4: Results on flying: (a) depicts PogoX in flying and disturbance injection; (b) compares the quality of trajectory tracking in $x-z$ plane; and (c) demonstrates how DDPC responds to disturbances in both height and leg angle tracking.
  • Figure 5: Results on periodic hopping: trajectories without reference-steering (green) and with reference-steering (blue).

Theorems & Definitions (5)

  • Definition 1
  • Definition 2
  • Theorem III.1: Williem's Fundamental Lemma
  • Definition 3
  • Theorem III.2