Robotics · Intermediate · 12-24 hours
Manipulation
Perception and control for grasping, moving, assembling, and using objects.
Direct answer
What is Manipulation?
Perception and control for grasping, moving, assembling, and using objects.
Definition and scope
Perception and control for grasping, moving, assembling, and using objects.
Systems estimate object and robot state, plan contacts and motion, and execute feedback control under uncertainty.
Why it matters
Manipulation is central to useful robots in homes, factories, laboratories, and logistics.
How it works
Systems estimate object and robot state, plan contacts and motion, and execute feedback control under uncertainty.
Beginner learning path
Learn coordinate frames, kinematics, grasping, motion planning, and visual feedback.
Recommended next topics
Primary sources
Key papers
RT-1: Robotics Transformer for Real-World Control at Scale
RT-1 trains one transformer policy on a large multi-task dataset of real robot demonstrations.
RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control
RT-2 co-trains vision-language models on web and robot data so semantic knowledge can influence actions.
OpenVLA: An Open-Source Vision-Language-Action Model
OpenVLA is an open 7B-parameter VLA trained on the Open X-Embodiment dataset.
Research ecosystem
Organizations working in this area
Organization
Figure AI
General-purpose humanoid robots and onboard VLA systems
View profile →Organization
Covariant
AI robotics for warehouse manipulation
View profile →Organization
Physical Intelligence
Generalist robot foundation models
View profile →Common questions
Frequently asked questions
What is Manipulation?
Perception and control for grasping, moving, assembling, and using objects.
Why does Manipulation matter for Physical AI?
Manipulation is central to useful robots in homes, factories, laboratories, and logistics.
How should a beginner learn Manipulation?
Learn coordinate frames, kinematics, grasping, motion planning, and visual feedback.