Research Analyzer
← Back ICRA 2026

V-MORALS: Visual Morse Graph-Aided Estimation of Regions of Attraction in a Learned Latent Space

Faiz Aladin, Ashwin Balasubramanian, Lars Lindemann, Daniel Seita

PDF

AI summary

Key figure (auto-extracted from paper)
V-MORALS enables accurate safety analysis and region-of-attraction estimation for robotic systems using only camera images, eliminating the need for explicit state measurements.
Reachability analysis Regions of Attraction Visual control Latent dynamics Morse graphs Safety verification

Problem

Current reachability and safety analysis methods typically require full state information or known system dynamics, making them inapplicable to high-dimensional robotic systems that are only observable through cameras.

Approach

The method compresses sequences of preprocessed images into a low-dimensional latent space using a 3D convolutional autoencoder, learns the underlying dynamics there, and constructs Morse Graphs to compute Regions of Attraction.

Key results

  • Extends MORALS to operate under partial observability using only image sequences
  • Introduces a spatiotemporal encoding pipeline with binary masking and 3D convolutional autoencoders
  • Generates accurate Morse Graphs and Regions of Attraction across four standard control benchmarks
  • Demonstrates reliable long-term outcome prediction (success/failure) without explicit state knowledge

Why it matters

Enables formal safety verification for complex robots operating in real-world environments where only visual sensors are available.

Abstract

Reachability analysis has become increasingly im- portant in robotics to distinguish safe from unsafe states. Unfortunately, existing reachability and safety analysis methods often fall short, as they typically require known system dynam- ics or large datasets to estimate accurate system models, are computationally expensive, and assume full state information. A recent method, called MORALS, aims to address these shortcomings by using topological tools to estimate Regions of Attraction (ROA) in a low-dimensional latent space. However, MORALS still relies on full state knowledge and has not been studied when only sensor measurements are available. This paper presents Visual Morse Graph-Aided Estimation of Regions of Attraction in a Learned Latent Space (V- MORALS). V-MORALS takes in a dataset of image-based trajectories of a system under a given controller, and learns a latent space for reachability analysis. Using this learned latent space, our method is able to generate well-defined Morse Graphs, from which we can compute ROAs for various systems and controllers. V-MORALS provides capabilities similar to the original MORALS architecture without relying on state knowledge, and using only high-level sensor data. Our project website is at: https://v-morals.onrender.com.

Index terms

Robot Safety AI-Based Methods Visual Learning

Related papers