1) The humanoid upper body (arms, hands, head, sensors) performs manipulation and interaction. 2) A wheeled mobile base (various layouts: differential, omnidirectional, self-balancing two-wheel) provides driving and positioning. 3) Navigation (SLAM, path planning, obstacle avoidance) moves the robot around a flat indoor space. 4) Coordination of base and arms (whole-body / mobile manipulation) lets the robot drive and manipulate objects at the same time.
Bipedal humanoids are expensive, complex, energy-hungry and prone to falling. A wheeled humanoid delivers human-like upper-body dexterity with simpler, stable and cheaper mobility - at the cost of stairs and rough terrain.
Torso, arms, hands and head with sensors - performing human-like manipulation and interaction.
A wheeled chassis (differential, omni or self-balancing) providing mobility instead of legs.
Official
SLAM, path planning and obstacle avoidance for indoor movement.
Official
Whole-body control coordinating base driving with arm motion (mobile/whole-body manipulation).
Official
Wheels limit the robot to flat indoor surfaces; thresholds, stairs and unevenness are a barrier.
Extending arms and lifting payload can risk tipping a narrow base.
A wheeled form is perceived differently from a bipedal humanoid; navigating among people requires care.
SoftBank Pepper popularizes a wheeled service humanoid for human interaction.
1X EVE shows a wheeled humanoid for work tasks as a pragmatic alternative to bipedal designs.
Time complexity: Nie dotyczy (form-factor sprzetowy, nie algorytm). Space complexity: Nie dotyczy.
Unlike bipedal humanoids, the wheeled base removes the costly balance/gait problem; compute is dominated by perception, navigation and mobile-manipulation coordination.
Differential, omnidirectional (omni) or self-balancing two-wheel drive.
The number of arm/hand DOF determines manipulation dexterity.
A fixed or telescoping torso height affects working reach.
This describes the robot's control stack; the form factor itself is not a compute paradigm.
Perception, navigation and arm control can run in parallel; whole-body coordination ties them together.
A robot design paradigm - not tied to a specific accelerator type; compute depends on the perception/control stack.
