AI can choose to lift a cup. Motors, sensors, control loops and safety systems must make that choice physically correct.
A humanoid robot does not turn an AI instruction into movement in one step. The decision must pass through a tightly packed electromechanical system that senses motion, controls force, manages power and responds to faults.
The Robot Report’s explainer presents this as an engineering problem rather than a report of a named robot entering production or paid deployment. Its central point is simple: AI can decide, but the joint still has to execute.
The decision is not the movement
A higher-level AI system may determine that a robot should pick up an object, turn a handle or hand someone a glass. That instruction does not directly move a motor. Each joint must continuously determine where it is, how it is moving and how much force it is applying.
Position, current and torque information feed a closed control loop. The system compares commanded movement with actual movement, then adjusts motor behavior. In plain English, the robot does not send an arm toward a coordinate and assume that the physical world will remain unchanged. It measures what is happening and corrects the motion as it unfolds.
Consider the apparently ordinary task of handing someone a cup of tea. The robot must know the position of its joints, control the torque applied as the arm moves and respond to changing physical conditions. The AI may select the action, but the joint-level electronics and control system must turn that intention into controlled movement.
That distinction matters because perception and planning solve different problems from actuation. Recognizing a cup is not the same as moving an arm toward it, regulating the grip or responding to contact.
Every joint carries a payload
A fixed industrial robot can leave much of its equipment outside the moving structure. It may be bolted to a factory floor, use thick copper cables and house motor drives in control cabinets. The system can also accommodate substantial cooling infrastructure if necessary.
A humanoid with 30 or more degrees of freedom has far tighter constraints. Every additional joint introduces another combination of motors, power electronics, position and current sensing, control and safety functions. All of that hardware has to travel with the robot.
The consequences accumulate. Every gram contributes to the energy required to move the machine. Every millimeter occupied by electronics competes for space inside an arm, leg, hand or finger. The problem is particularly acute in a dexterous hand, where multiple motors, sensors and control functions must fit into very small spaces.
That makes the joint more than a motor attached to a limb. The Robot Report describes it as an integrated electromechanical subsystem combining sensing, motor control, power management, thermal performance and functional safety. It calls the joint a “critical interface between AI and the physical world.”
This is why improving the AI alone cannot remove the physical bottleneck. The robot still needs compact hardware capable of surviving and controlling the forces generated by movement.
Why voltage becomes a mobility issue
Power architecture shows how closely electrical and mechanical design are linked.
The Robot Report describes more advanced mobile robots and humanoids moving toward 48-volt DC architectures. For the same delivered power, raising distribution voltage from 12 volts to 48 volts reduces the required current by a factor of four. Designers can then use thinner wiring, reducing copper weight throughout the machine.
The electrical benefit compounds. Because resistive losses scale with the square of current, reducing current by four reduces wiring losses by a factor of 16. In a humanoid, lower wiring weight reduces the energy needed for movement, while lower thermal loads make it easier to package electronics into joints without elaborate cooling systems.
Higher voltage also creates another design demand. During rapid deceleration, motors can behave as generators and produce transient voltages considerably higher than the nominal DC bus voltage. The electronics controlling the joint must withstand those events without damage or loss of control.
So a power decision is also a packaging, heat and reliability decision. Lower current can reduce wiring weight and losses, but the motor-control electronics must handle more demanding electrical events. Power, thermal performance, weight and reliability affect one another.
Feedback turns contact into controlled action
Touch and force feedback extend the control problem beyond simple positioning.
When a person grips a tool or handles a fragile object, the movement changes in response to what the fingers feel. A robot with force and touch sensing in its fingers or other contact surfaces can similarly adjust its movement as contact occurs instead of replaying a rigid path.
The Robot Report identifies human-like dexterity, particularly in the hand, as a significant robotics challenge. Different joints also impose different sensing and packaging requirements. A shoulder and a finger do not have the same mechanical demands. Large motors can create electromagnetic interference that sensing systems must tolerate, while small joints leave little room for precise feedback hardware.
There is unlikely to be one sensor arrangement that suits every joint. Reliable feedback is part of the control loop that lets a robot respond when the physical world differs from its plan.
Safety has to live inside the control system
Industrial automation often manages risk by separating people and machines. A humanoid designed to collaborate with a factory worker or assist with tasks in a home cannot depend on that separation.
Functional safety therefore becomes part of the robot itself. The control system must monitor operating conditions, detect faults or anomalous behavior and quickly move an affected subsystem into an appropriate safe state when necessary.
The source describes safety as something that should be designed into the system from the beginning rather than layered onto a finished machine. For a humanoid operating near people, that means safety monitoring and fault response belong inside the control architecture alongside sensing and motor control.
The practical chain is therefore: an AI system selects an action; joint electronics measure position, current and torque; closed-loop controls adjust motor behavior; touch and force feedback can modify movement during contact; and safety functions monitor for conditions requiring a controlled response.
For anyone evaluating a humanoid demonstration, the useful next step is to look beyond what the robot appears to understand. Ask how its joints sense movement and force, how its power system handles heat and electrical transients, and how the machine reaches a safe state when conditions change. The AI may choose the action. The integrated joint determines whether that choice becomes dependable physical movement.