Implicit Contact Diffuser: Sequential Contact Reasoning With Latent Point Cloud Diffusion

Long-horizon contact-rich manipulation has long been a challenging problem, as it requires reasoning over both discrete contact modes and continuous object motion. We introduce Implicit Contact Diffuser (ICD), a diffusion-based model that generates a sequence of neural descriptors that specify a series of contact relationships between the object and the environment. This sequence is then used as guidance for an MPC method to accomplish a given task. The key advantage of this approach is that the latent descriptors provide more taskrelevant guidance to MPC, helping to avoid local minima for contact-rich manipulation tasks. Our experiments demonstrate that ICD outperforms baselines on complex, long-horizon, contact-rich manipulation tasks, such as cable routing and notebook folding. Additionally, our experiments also indicate that ICD can generalize a target contact relationship to a different environment. More visualizations can be found on our website https://implicit-contact-diffuser.github.io.

Authors

• Zixuan Huang
• Yinong He
• Yating Lin
• Dmitry Berenson

Keywords

• Point cloud compression
• Visualization
• Routing
• Cognition
• Robotics and automation
• Point Cloud
• Notebook
• Manipulation Tasks
• Model Predictive Control
• Continuous Motion
• Contact Relationship
• Local Features
• Error Model
• Reachable
• Diffusion Model
• Latent Model
• Neural Representations
• Object Parts
• Variational Autoencoder
• Latent Vector
• Object Point
• Contact Interaction
• Object Pose
• Regularization Loss
• Implicit Representation
• Object Point Cloud
• Model Predictive Control Framework
• Deformable Objects
• Signed Distance Function
• Perceptual Errors
• Neural Field
• Robot Motion
• Contact Patch
• Point Cloud Generation
• Hybrid System