ESP: Energy-Score Policy for One-Step Multimodal Action Generation

Key points

• We propose ESP (Energy-Score Policy), a teacher-free approach that maps policy context and noise directly to an action chunk in a single network evaluation.1
• This provides a principled objective for learning multimodal action distributions without iterative sampling, with exact recovery at the population optimum when the model can represent the target distribution.1
• Experiments on both simulation and real-world manipulation tasks demonstrate competitive task success with substantially lower action-generation latency than the flow matching baseline.1
• These results support direct distributional learning as an efficient alternative to iterative generative robot policies.1

Sources (1)

• [1]ESP: Energy-Score Policy for One-Step Multimodal Action Generation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 03:34 AM

  Generative action models based on diffusion and flow matching have been increasingly adopted in vision-language-action (VLA) policies for their ability to capture diverse behaviors, including multiple valid action sequences under the same observation and instruction.

  We propose ESP (Energy-Score Policy), a teacher-free approach that maps policy context and noise directly to an action chunk in a single network evaluation.

Extractive summary: sentences quoted from the sources.