In traditional robotics, actuators are often placed directly at the joints they power. While simple to model, this "serial" approach creates a fundamental performance ceiling: every motor added to a limb becomes a heavy burden for the motors closer to the base to carry. This distal mass increases rotational inertia, requiring more energy to move, slowing down response times, and increasing the kinetic energy—and thus the danger—of impacts.
A growing body of research focuses on proximal actuator relocation, a design strategy that moves heavy motors and gearboxes away from the extremities and toward the robot's base or center of mass. By decoupling the location of the power source from the joint it manipulates, researchers are creating robots that are faster, safer, and more energy-efficient.
The Physics of Distal Mass
The primary motivation for relocating actuators is the reduction of rotational inertia. As Won Dong Shin et al. observe in their study of avian-inspired robots, birds concentrate their muscle mass near the hip joint. This proximal placement minimizes the energy required to swing the leg during locomotion. Conversely, placing heavy components at the extremities increases the metabolic cost of movement—a critical factor in wearable robotics. Guoping Zhao et al. and Shuo Wang and Ziran Zhang emphasize that for exoskeletons, reducing distal mass is essential to ensure that the assistance provided by the device isn't negated by the energy the user spends just to move the device's own weight.
Beyond efficiency, proximal actuation enables a rare combination of speed and safety. Simon Guist et al. demonstrate that by moving actuators to the base of a robotic arm, they can achieve professional-level table tennis speeds while maintaining low impact forces. This effectively decouples the robot's operational speed from its kinetic energy hazards, making high-speed robots safer for human interaction.
Mechanisms of Remote Power Transmission
Relocating an actuator requires a way to transmit torque across a distance. Researchers generally employ three mechanical strategies to achieve this:
- Cable and Tendon Drives: By using high-strength fibers or Bowden cables, designers can place motors at the waist or hip and pull on joints remotely. Vassilios Tsounis and Jemin Hwangbo utilized this in the Capler-Leg, achieving a 96.5% energy recuperation rate by replacing heavy gearboxes with a high-performance cable-pulley system.
- Parallel Linkages: Instead of a single chain of bones, robots like the Duke Humanoid (Boxi Xia et al.) and LEVA (Marco Arnold et al.) use parallel four-bar or parallelogram linkages. These allow motors for the knee or ankle to be mounted coaxially at the hip, significantly reducing the leg's swing inertia.
- Quasi-Direct Drive (QDD): Often paired with proximal relocation, QDD actuators use low gear ratios to maintain "transparency." As Shuo Wang and Ziran Zhang explain, this allows the robot to be highly back-driveable, meaning it offers little resistance when the user moves naturally, only providing torque when specifically commanded.
Overcoming Modeling Complexity
While proximal actuation offers mechanical advantages, it introduces significant mathematical hurdles. Traditional robot description formats (like URDF) are designed for "open-chain" serial robots. Relocating actuators via linkages creates kinematic loops—closed circuits of motion where the movement of one link is constrained by several others.
To address this, Jean-Jacques Slotine et al. developed URDF+, an extension that allows for the automated parsing of these complex loops. Similarly, Ludovic De Matteïs and Justin Carpentier have proposed motion generation frameworks that handle closed-loop kinematics directly. These tools allow researchers to leverage the full mechanical advantage of parallel structures without relying on "serial approximations" that can lead to control failures in high-performance maneuvers.
Power Density and Active Mass
A final benefit of this research direction is the optimization of the "active weight ratio." Vassilios Tsounis and Jemin Hwangbo argue that many robots are dominated by "passive weight"—housings, bolts, and structural supports. By clustering actuators at the base, designers can make the power-producing elements (rotors and stators) constitute the majority of the limb's weight.
This clustering also enables torque sharing. In the SpaceHopper robot (Alexander Spiridonov et al.), a parallel differential drive allows multiple motors to combine their power for high-load phases like jumping, then switch to lower-load orientation adjustments, maximizing the utility of every gram of motor weight.