humanoidsdata.com

Search

Search datasets, articles, and glossary terms for humanoids and embodied AI.

← All glossary terms

Hardware & control

Mobile manipulation

Mobile manipulation is the coordinated use of locomotion and manipulation by a robot that combines a mobile base or body with one or more arms or other manipulators. Mobility expands the robot's reachable workspace, while the manipulator grasps, pushes, operates tools or makes contact; the base and arm may move separately or at the same time.

Also known as: mobile manipulator, mobile manipulators

Updated

Mobility changes the reachable workspace

Modern Robotics defines a mobile manipulator as a mobile base equipped with one or more robot arms. A fixed arm can manipulate only within its installed workspace. A mobile manipulator can reposition the base, carry an object between locations and approach several stations with the same arm.

Wheels are common, but mobility can also come from tracks or legs. A humanoid that walks to a shelf and picks an object is performing mobile manipulation even though its limbs and base are part of one articulated body.

The base and manipulator must be coordinated

A simple system can drive to a location, stop and then move its arm. Harder tasks require simultaneous motion. Khatib's mobile-manipulation framework treats the mobile base as a coarse, slower mechanism and the arm as a faster, more accurate manipulator, then coordinates both around the end-effector task.

Coordination matters when opening a large door, carrying an object while walking or following a surface that extends beyond the arm's stationary reach. The controller must allocate motion between locomotion and arm joints while respecting collisions, joint limits, balance or wheel constraints and contact forces.

Mobile manipulation is not one autonomy level

The term describes a physical capability, not how decisions are made. A mobile manipulator may be autonomous, teleoperated or use variable autonomy. The wheeled mobile-manipulation survey covers applications ranging from inspection to logistics resupply and assembly under both autonomous and teleoperated control.

For training data, a mobile-manipulation episode should preserve base pose and velocity, arm and hand state, end-effector targets, sensor observations, actions, contacts and the timing between navigation and manipulation. Recording only the hand trajectory loses the body movement that made the task reachable.

Sources