Subconscious Robotic Imitation Learning

TL;DR

This paper introduces Subconscious Robotic Imitation Learning (SRIL), a novel approach inspired by human subconscious processing that significantly accelerates robotic task execution by reducing model inference delays.

Contribution

The paper proposes SRIL, combining cognitive offloading, historical action chunking, and pattern augmented learning to improve real-time responsiveness in robotic imitation learning.

Findings

SRIL achieves 100-200% faster execution than state-of-the-art policies.

SRIL maintains higher success rates in dual-arm manipulation tasks.

The approach effectively reduces inference delays in robotic learning.

Abstract

Although robotic imitation learning (RIL) is promising for embodied intelligent robots, existing RIL approaches rely on computationally intensive multi-model trajectory predictions, resulting in slow execution and limited real-time responsiveness. Instead, human beings subconscious can constantly process and store vast amounts of information from their experiences, perceptions, and learning, allowing them to fulfill complex actions such as riding a bike, without consciously thinking about each. Inspired by this phenomenon in action neurology, we introduced subconscious robotic imitation learning (SRIL), wherein cognitive offloading was combined with historical action chunkings to reduce delays caused by model inferences, thereby accelerating task execution. This process was further enhanced by subconscious downsampling and pattern augmented learning policy wherein intent-rich information was addressed with quantized sampling techniques to improve manipulation efficiency. Experimental results demonstrated that execution speeds of the SRIL were 100\% to 200\% faster over SOTA policies for comprehensive dual-arm tasks, with consistently higher success rates.