Object Rearrangement Using Learned Implicit Collision Functions

Robotic object rearrangement combines the skills of picking and placing objects. When object models are unavailable, typical collision-checking models may be unable to predict collisions in partial point clouds with occlusions, making generation of collision-free grasping or placement trajectories challenging. We propose a learned collision model that accepts scene and query object point clouds and predicts collisions for 6DOF object poses within the scene. We train the model on a synthetic set of 1 million scene/object point cloud pairs and 2 billion collision queries. We leverage the learned collision model as part of a model predictive path integral (MPPI) policy in a tabletop rearrangement task and show that the policy can plan collision-free grasps and placements for objects unseen in training in both simulated and physical cluttered scenes with a Franka Panda robot. The learned model outperforms both traditional pipelines and learned ablations by 9. 8% in accuracy on a dataset of simulated collision queries and is 75x faster than the best-performing baseline. Videos and supplementary material are available at https://research.nvidia.com/publication/2021-03_Object-Rearrangement-Using.

Authors

• Michael Danielczuk
• Arsalan Mousavian
• Clemens Eppner
• Dieter Fox

Keywords

• Training
• Time-frequency analysis
• Process control
• Predictive models
• Trajectory
• Planning
• Task analysis
• Object Rearrangement
• Learning Models
• Point Cloud
• Collision Detection
• Object Pose
• Scene Point
• Simulated Scene
• Object Point Cloud
• Neural Network
• Multilayer Perceptron
• Object Features
• Target Object
• Path Planning
• 3D Mesh
• Joint Space
• End-effector
• Single Pass
• Objects In The Scene
• Latent Vector
• Average Precision Score
• Point Cloud Data
• Collision-free Trajectory
• Raw Point Cloud
• Signed Distance Function
• Mesh Model
• Placement Position
• Object Geometry
• Linear Trajectory