Fuzzy Encoding-Decoding to Improve Spiking Q-Learning Performance in Autonomous Driving

TL;DR

This paper presents a fuzzy encoder-decoder architecture that significantly improves the performance of spiking Q-networks in autonomous driving by enhancing information encoding and value function representation.

Contribution

It introduces trainable fuzzy membership functions and a lightweight neural decoder to address information loss and limited capacity in spiking reinforcement learning.

Findings

Substantially improved decision-making accuracy on HighwayEnv benchmark.

Close performance gap between spiking and non-spiking multi-modal Q-networks.

Demonstrated potential for real-time autonomous driving with spiking neural networks.

Abstract

This paper develops an end-to-end fuzzy encoder-decoder architecture for enhancing vision-based multi-modal deep spiking Q-networks in autonomous driving. The method addresses two core limitations of spiking reinforcement learning: information loss stemming from the conversion of dense visual inputs into sparse spike trains, and the limited representational capacity of spike-based value functions, which often yields weakly discriminative Q-value estimates. The encoder introduces trainable fuzzy membership functions to generate expressive, population-based spike representations, and the decoder uses a lightweight neural decoder to reconstruct continuous Q-values from spiking outputs. Experiments on the HighwayEnv benchmark show that the proposed architecture substantially improves decision-making accuracy and closes the performance gap between spiking and non-spiking multi-modal Q-networks. The results highlight the potential of this framework for efficient and real-time autonomous driving with spiking neural networks.