Global robotics landscape · Family 03
Humanoids and mobile manipulators
A grounded guide to machines that combine locomotion, perception and manipulation—separating morphology, task capability and sustained operations.
FUURAA thesis
Human-like form may improve access to human infrastructure, but operational value depends on task-level reliability and safe recovery.
How to use this page
Move from what it is to how to judge it—without being misled by one specification or demo.
Separate embodiment, sensing, control, tooling and operations.
Open in-depth topic →02WorkRead value through the complete task and environment, not movement alone.
Open in-depth topic →03EvidenceTest claims with repeatable metrics, failures and human intervention.
Open in-depth topic →04BoundaryKeep standards, regulation, site and research-stage limits visible.
Open in-depth topic →System anatomy
A working robot is the result of several engineering layers holding together.
Each layer explains its job and the signal most worth verifying next.
Locomotion and balance
Legs or wheels trade terrain access, energy use, stability, speed and fall consequences differently.
Open in-depth topic →02Whole-body manipulation
Hands, arms, torso and base must coordinate contact, reach, force and collision avoidance around clutter.
Open in-depth topic →03Perception and task policy
Scene understanding, object state, language instructions and action planning must stay grounded in recoverable physical states.
Open in-depth topic →04Teleoperation and fleet operations
Human assistance, remote supervision, charging, updates and incident review may be essential parts of the product.
Open in-depth topic →Real-work map
Form factor is an entry point; the complete workflow is the unit of value.
These are not capability guarantees; they frame the task boundaries and evidence a reader should seek.
Material movement plus handling
Moving totes, components or tools tests navigation, docking, grasping and placement as one chain.
Open in-depth topic →02Machine and facility support
Opening doors, pressing controls and tending equipment are attractive because they use existing human infrastructure.
Open in-depth topic →03Inspection with intervention
A mobile manipulator can inspect first and physically respond, but intervention raises contact and authority risks.
Open in-depth topic →04Research and data collection
General platforms support policy learning and embodiment studies; research progress should not be presented as product availability.
Open in-depth topic →Evaluation checklist
Five questions turn product claims into testable deployment judgements.
Procurement, replication, pilot design and policy review should record success, failure and human cost together.
- 01
Task definition
Specify start state, completion criteria, tolerated damage, time limit and allowed human assistance.
Open in-depth topic → - 02
Reliability distribution
Report repeated trials across object, lighting, layout and operator variation, including failures.
Open in-depth topic → - 03
Intervention burden
Measure remote assists, physical resets, prompt edits and recovery time per productive hour.
Open in-depth topic → - 04
Energy and duty cycle
Include active work, waiting, charging, thermal limits and battery replacement logistics.
Open in-depth topic → - 05
Safe degraded behaviour
Test loss of perception, communications, balance, tool state and power under controlled conditions.
Open in-depth topic →
Scope boundaries
Make explicit what is unknown and what cannot be generalised.
- 01
A choreographed demonstration cannot establish autonomous operation, availability or economic value.
Open in-depth topic → - 02
Human-like hands and body proportions do not imply human-level judgement, dexterity or safe social behaviour.
Open in-depth topic → - 03
Factory pilots should not be generalized to homes or public spaces, where users and environments are less controlled.
Open in-depth topic →
Sources and evidence status
Return to the original record before deciding how far a conclusion can travel.
Each source carries a publication or review date, evidence status and the conclusion it cannot support alone.
Grasping, manipulation and contact-safety measurement
Develops performance metrics and test methods for dexterity, control and contact safety rather than relying on demonstration videos.
Standard test methods for response robots
Provides repeatable ways to quantify mobility, manipulation, sensing, endurance, communications, operator proficiency, logistics and safety.