A model-based framework for learning transparent swarm behaviors

TL;DR

This paper introduces a model-based framework for designing transparent, verifiable swarm behaviors by extracting and utilizing models of local and global interactions, enabling effective controller synthesis and behavior inspection.

Contribution

The paper presents a novel framework that automatically derives models from simulation data to optimize swarm behaviors and enhance their interpretability and verifiability.

Findings

Framework successfully applied to aggregation and foraging tasks

Models enable understanding and inspection of swarm behaviors

Verification checks identify potential issues in swarm performance

Abstract

This paper proposes a model-based framework to automatically and efficiently design understandable and verifiable behaviors for swarms of robots. The framework is based on the automatic extraction of two distinct models: 1) a neural network model trained to estimate the relationship between the robots' sensor readings and the global performance of the swarm, and 2) a probabilistic state transition model that explicitly models the local state transitions (i.e., transitions in observations from the perspective of a single robot in the swarm) given a policy. The models can be trained from a data set of simulated runs featuring random policies. The first model is used to automatically extract a set of local states that are expected to maximize the global performance. These local states are referred to as desired local states. The second model is used to optimize a stochastic policy so as to increase the probability that the robots in the swarm observe one of the desired local states. Following these steps, the framework proposed in this paper can efficiently lead to effective controllers. This is tested on four case studies, featuring aggregation and foraging tasks. Importantly, thanks to the models, the framework allows us to understand and inspect a swarm's behavior. To this end, we propose verification checks to identify some potential issues that may prevent the swarm from achieving the desired global objective. In addition, we explore how the framework can be used in combination with a "standard" evolutionary robotics strategy (i.e., where performance is measured via simulation), or with online learning.