EquiContact: A Hierarchical SE(3) Vision-to-Force Equivariant Policy for Spatially Generalizable Contact-rich Tasks

2025/07/15 by Joohwan Seo, Arvind Kruthiventy, Seo, Joohwan +12 · 2 citations
Engineering · Neuroscience · #EEG and Brain-Computer Interfaces #FOS: Computer and information sciences #Robot Manipulation and Learning #Robotics (cs.RO) #Tactile and Sensory Interactions

paper · pdf · doi:10.48550/arxiv.2507.10961

openalex publication_date 2025/07/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper presents a framework for learning vision-based robotic policies for contact-rich manipulation tasks that generalize spatially across task configurations. We focus on achieving robust spatial generalization of the policy for the peg-in-hole (PiH) task trained from a small number of demonstrations. We propose EquiContact, a hierarchical policy composed of a high-level vision planner (Diffusion Equivariant Descriptor Field, Diff-EDF) and a novel low-level compliant visuomotor policy (Geometric Compliant ACT, G-CompACT). G-CompACT operates using only localized observations (geometrically consistent error vectors (GCEV), force-torque readings, and wrist-mounted RGB images) and produces actions defined in the end-effector frame. Through these design choices, we show that the entire EquiContact pipeline is SE(3)-equivariant, from perception to force control. We also outline three key components for spatially generalizable contact-rich policies: compliance, localized policies, and induced equivariance. Real-world experiments on PiH, screwing, and surface wiping tasks demonstrate a near-perfect success rate and robust generalization to unseen spatial configurations, validating the proposed framework and principles. The experimental videos and more details can be found on the project website: https://equicontact.github.io/EquiContact-website/

Citations

• A Practical Guide for Incorporating Symmetry in Diffusion Policy
• Tool-as-Interface: Learning Robot Policies from Observing Human Tool Use
• SE(3)-Equivariant Robot Learning and Control: A Tutorial Survey
• Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation
• Hierarchical Equivariant Policy via Frame Transfer
• FoAR: Force-Aware Reactive Policy for Contact-Rich Robotic Manipulation
• SE(3) Equivariant Ray Embeddings for Implicit Multi-View Depth Estimation
• ET-SEED: Efficient Trajectory-Level SE(3) Equivariant Diffusion Policy
• π₀: A Vision-Language-Action Flow Model for General Robot Control
• One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation
• Learning from 10 Demos: Generalisable and Sample-Efficient Policy Learning with Oriented Affordance Frames
• The Ingredients for Robotic Diffusion Transformers
• ActionFlow: Equivariant, Accurate, and Efficient Policies with Spatially Symmetric Flow Matching
• EquiBot: SIM(3)-Equivariant Diffusion Policy for Generalizable and Data Efficient Learning
• Equivariant Diffusion Policy
• Learning Variable Compliance Control From a Few Demonstrations for Bimanual Robot with Haptic Feedback Teleoperation System
• OpenVLA: An Open-Source Vision-Language-Action Model
• Neural Isometries: Taming Transformations for Equivariant ML
• Riemannian Flow Matching Policy for Robot Motion Learning
• A Comparison Between Lie Group- and Lie Algebra- Based Potential Functions for Geometric Impedance Control
• Fourier Transporter: Bi-Equivariant Robotic Manipulation in 3D
• Diffusion-EDFs: Bi-equivariant Denoising Generative Modeling on SE(3) for Visual Robotic Manipulation
• Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware
• Sigmoid Loss for Language Image Pre-Training
• Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
• Diffusion policy: Visuomotor policy learning via action diffusion
• Scalable Diffusion Models with Transformers
• Behavior Transformers: Cloning k modes with one stone
• Neural Descriptor Fields: SE(3)-Equivariant Object Representations for Manipulation
• CLIPort: What and Where Pathways for Robotic Manipulation
• Learning Transferable Visual Models From Natural Language Supervision
• Transporter Networks: Rearranging the Visual World for Robotic Manipulation
• Object-Centric Learning with Slot Attention
• FiLM: Visual Reasoning with a General Conditioning Layer
• FastUMI: A Scalable and Hardware-Independent Universal Manipulation Interface with Dataset

Cited by

• Symmetry-Aware Steering of Equivariant Diffusion Policies: Benefits and Limits
• Partially Equivariant Reinforcement Learning in Symmetry-Breaking Environments
• Geometric Formulation of Unified Force-Impedance Control on SE(3) for Robotic Manipulators

Related

• EquivAct: SIM(3)-Equivariant Visuomotor Policies beyond Rigid Object Manipulation
• Admittance Visuomotor Policy Learning for General-Purpose Contact-Rich Manipulations
• Teaching contact-rich tasks from visual demonstrations by constraint extraction
• Making Sense of Vision and Touch: Self-Supervised Learning of Multimodal Representations for Contact-Rich Tasks
• Multi-Modal Fusion in Contact-Rich Precise Tasks via Hierarchical Policy Learning