skills.rest
Gonglitian
GonglitianCOMMUNITY
Explore all skills by this author →
GitHub

paper_rob__3d_diffusion_policy

Generate robot action trajectories from sparse 3D point clouds using a diffusion policy.

2|Updated Mar 20, 2026
Search Tags:#robotics#diffusion#visuomotor#pointcloud#imitation-learning#pointnet#FiLM
One-click install
npx skills add https://github.com/Gonglitian/agent-skills --skill paper-rob-3d-diffusion-policy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: paper_rob__3d_diffusion_policy
Source: https://github.com/Gonglitian/agent-skills/tree/main/skills/papers/paper_rob__3d_diffusion_policy
Command: npx skills add https://github.com/Gonglitian/agent-skills --skill paper-rob-3d-diffusion-policy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

DP3 enables learning and inference of visuomotor actions from sparse 3D point clouds for robotics, reducing reliance on dense sensor data or extensive demonstrations.

Core Features & Use Cases

  • DP3 policy overview: A diffusion-based policy that conditions action generation on compact 3D features extracted from sparse point clouds and proprioceptive state.
  • Encoder & diffusion architecture: PointNet-based encoder (DP3Encoder) combined with a FiLM-conditioned 1D UNet to produce denoised action sequences.
  • Use Case: Learn visuomotor policies for tasks like object manipulation with few demonstrations in both simulation and real robots.

Quick Start

Provide a point_cloud and agent_pos observation to the DP3 policy to generate a sequence of actions.

Frequently Asked Questions about paper_rob__3d_diffusion_policy

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I learn visuomotor policies from sparse 3D point clouds with limited demonstrations?▼

Yes, you can train imitation learning policies for real robots using a 3D diffusion policy that infers executable action trajectories from sparse point cloud observations and agent positions, applied directly in real-world manipulation settings.

Can I use a 3D diffusion policy for real robot imitation learning tasks?▼

Yes, you can train imitation learning policies for real robots using a 3D diffusion policy that infers executable action trajectories from sparse point cloud observations and agent positions, applied directly in real-world manipulation settings.

What inputs do I need to generate robot actions with a 3D diffusion policy?▼

To generate robot actions, provide point_cloud and agent_pos observations to the diffusion policy, which uses a PointNet-based encoder and FiLM-conditioned 1D UNet to produce denoised action sequences.

How does the PointNet encoder and FiLM-conditioned UNet architecture work for diffusion policies?▼

The architecture uses a PointNet-based encoder to extract compact 3D features from sparse point clouds, which are fed into a FiLM-conditioned 1D UNet diffusion model to produce denoised action sequences.

Does 3D diffusion policy reduce the need for dense sensor data in robotic imitation tasks?▼

Yes, 3D diffusion policy reduces reliance on dense sensor data by enabling learning and inference of visuomotor actions from sparse 3D point clouds across simulated and real environments.

Related Skills

Explore More Education & Research Skills→
NVIDIA-TAONVIDIA-TAO
83

tao-train-pointpillars

Trains, evaluates, exports, and prunes PyTorch-based PointPillars 3D detection models from LiDAR point clouds.

Official
Advanced
zjunlpzjunlp
1.1k

alfworld-tool-user

Apply tools to target objects in simulated ALFWorld environments.

Official
Intermediate
melikhanmutlumelikhanmutlu

computer-vision-expert

Design computer vision pipelines integrating YOLO26, SAM 3, and VLMs for edge deployment.

Community
Advanced
kngender5kngender5

image-to-3d-generation

Convert 2D images into 3D models using AI pipelines.

Community
Advanced