Brain-inspired Action Generation with Spiking Transformer Diffusion Policy Model

TL;DR

This work addresses generating coherent robot action trajectories with brain-inspired computation by integrating Spiking Neural Networks (SNNs), diffusion-policy learning, and Transformer architectures. It introduces STMDP, a Spiking Transformer-based diffusion policy, and a Spiking Modulate Decoder (SMD) that modulates cross-attention to guide action generation. Across four robotic manipulation tasks, STMDP demonstrates strong performance, including an 8% improvement on the Can task and a 75.4% score on PushT, with ablations confirming the benefit of the modulation block. The study highlights a promising brain-inspired direction for robotics by combining SNNs, diffusion models, and Transformer decoders, with data available via the BrainCog Embot platform.

Abstract

Spiking Neural Networks (SNNs) has the ability to extract spatio-temporal features due to their spiking sequence. While previous research has primarily foucus on the classification of image and reinforcement learning. In our paper, we put forward novel diffusion policy model based on Spiking Transformer Neural Networks and Denoising Diffusion Probabilistic Model (DDPM): Spiking Transformer Modulate Diffusion Policy Model (STMDP), a new brain-inspired model for generating robot action trajectories. In order to improve the performance of this model, we develop a novel decoder module: Spiking Modulate De coder (SMD), which replaces the traditional Decoder module within the Transformer architecture. Additionally, we explored the substitution of DDPM with Denoising Diffusion Implicit Models (DDIM) in our frame work. We conducted experiments across four robotic manipulation tasks and performed ablation studies on the modulate block. Our model consistently outperforms existing Transformer-based diffusion policy method. Especially in Can task, we achieved an improvement of 8%. The proposed STMDP method integrates SNNs, dffusion model and Transformer architecture, which offers new perspectives and promising directions for exploration in brain-inspired robotics.

Figures (4)

• Figure 1: The architect of spiking transformer modulate diffusion policy module.
• Figure 2: The architect of base block.(a) represents the architecture of Spiking Encoder;(b) represents the architecture of SMD.
• Figure 3: Tasks used for the experiment.
• Figure 4: The results of the ablation study.