PPGuide: Steering Diffusion Policies with Performance Predictive Guidance
2026-03-11 • Robotics
Robotics
AI summaryⓘ
The authors present PPGuide, a method to improve robot actions learned through diffusion policies by avoiding mistakes during execution. PPGuide uses a new way to figure out which parts of an action sequence lead to success or failure by analyzing the robot's past behavior without needing extra labeled data. Then, it trains a model to predict performance and guides the robot in real time to choose better actions. Tests on various robot tasks showed that PPGuide consistently helps robots perform better.
diffusion policiesrobotic manipulationaction sequencesexpert demonstrationsperformance predictionattention-based learningmultiple instance learningself-supervised learningpolicy guidanceinference time
Authors
Zixing Wang, Devesh K. Jha, Ahmed H. Qureshi, Diego Romeres
Abstract
Diffusion policies have shown to be very efficient at learning complex, multi-modal behaviors for robotic manipulation. However, errors in generated action sequences can compound over time which can potentially lead to failure. Some approaches mitigate this by augmenting datasets with expert demonstrations or learning predictive world models which might be computationally expensive. We introduce Performance Predictive Guidance (PPGuide), a lightweight, classifier-based framework that steers a pre-trained diffusion policy away from failure modes at inference time. PPGuide makes use of a novel self-supervised process: it uses attention-based multiple instance learning to automatically estimate which observation-action chunks from the policy's rollouts are relevant to success or failure. We then train a performance predictor on this self-labeled data. During inference, this predictor provides a real-time gradient to guide the policy toward more robust actions. We validated our proposed PPGuide across a diverse set of tasks from the Robomimic and MimicGen benchmarks, demonstrating consistent improvements in performance.