Robot control
Loco-manipulation
Definition
Loco-manipulation is the coordinated use of locomotion and physical object manipulation within one robot task. A humanoid may walk, change support contacts, reach, and apply forces while accounting for how those actions affect one another.
Also known as: Locomanipulation, Locomotion-manipulation
Updated
Locomotion and manipulation share the same body
Walking toward an object and moving it are not always separable phases. Carrying a bulky load changes a humanoid's mass distribution. Pushing a cart creates hand forces that can affect balance. Reaching beyond the current stance may require a new footstep. Loco-manipulation treats these effects as one coupled problem rather than sending unrelated commands to the legs and arms.
The coupling is especially important for a humanoid robot, because its hands and feet can all form or break contacts. A controller or planner may need to coordinate the object's motion, the robot's posture, and the sequence of those contacts.
It can be planned or learned
There is no single loco-manipulation algorithm. Wang and colleagues formulate dynamic humanoid loco-manipulation with multi-contact model predictive control. Their model includes foot-ground and hand-object contact modes, while treating the object's effect as an external force on the robot.
Learning-based systems can instead train a policy for longer sequences. The 2026 Humanoid Horizon preprint reports simulated and hardware demonstrations in which a humanoid navigates, grasps, transports, and places objects. Those experiments support the method on the tested tasks and robot; they do not establish general competence in unseen homes or workplaces.
Integration adds constraints
A successful grasp does not guarantee a feasible walking motion, and a stable gait does not guarantee useful object motion. Joint limits, self-collision, contact friction, payload uncertainty, and restricted sensing can all couple the two parts of the task. Errors may also compound over a long sequence as the robot changes stance and interacts with the object.
Loco-manipulation is therefore narrower than general whole-body control and broader than walking while holding a fixed pose. It describes tasks in which moving through the environment and manipulating something are both essential to the outcome.
Sources
Related terms
Whole-body control
Whole-body control coordinates a robot’s joints and contacts to satisfy several motion and force objectives together. In humanoids, it commonly combines balance, foot motion, hand tasks, and posture subject to physical constraints.
Bipedal locomotion
Bipedal locomotion is movement using two legs, with body motion coordinated through changing contacts between the feet and the environment. It includes walking and running.
Robotic manipulation
Robotic manipulation is the use of a robot to change an object's position, orientation, or state through physical interaction. It includes grasping and moving objects as well as actions such as pushing or carrying them without a grasp.
Model predictive control
Model predictive control repeatedly optimizes future actions using a system model, applies the next part of the solution, and replans from updated state information. It can account for objectives and constraints over a finite prediction horizon.