Engineering
Designing Series Elastic Actuators for Safe Human-Robot Interaction
Quick fact
Series elastic actuators were introduced by Gill Pratt and Matthew Williamson in 1995 at MIT, and they dramatically improved a robot's ability to estimate and control forces: by measuring spring deflection, the actuator can accurately sense force without needing a separate force sensor.
Why this is interesting
When a robot arm accidentally bumps into you, why does it hurt less if it has a spring inside? And how can a spring actually help the robot control its force more precisely?
Read the full explanation
Understanding Designing Series Elastic Actuators for Safe Human-Robot Interaction
Imagine a robot arm powered by a motor. In a conventional rigid actuator, the motor is directly connected to the joint via a gearbox, so the joint moves exactly as the motor commands. If something blocks the motion, the motor pushes against it with full force, which can be dangerous or damaging. A series elastic actuator (SEA) inserts a spring between the motor and the joint. Now, when the motor turns, it first compresses the spring, and the spring then pushes the joint. The joint's motion becomes a delayed, softened version of the motor's motion. This spring acts as a mechanical filter: it absorbs sudden impacts, smooths out torque spikes, and stores energy that can be released for agile movements. The key insight is that the spring deflection tells you the force being transmitted: force = spring stiffness × deflection. By measuring how much the spring compresses (using an encoder), the controller knows exactly how much force is applied at the joint. This enables remarkably precise force control, which is essential for tasks like gently shaking hands or applying a consistent force during polishing. So, an SEA makes a robot both safer and more force-aware, at the cost of some responsiveness and added complexity.
A deeper explanation
The mechanism that makes an SEA work is the intentional insertion of a compliant element—typically a spring—between the motor output and the load. The motor is still the source of power, but the spring decouples the motor's inertia from the load, reducing the effective inertia felt at the joint. This is crucial because a high-inertia motor can cause severe damage during unexpected collisions. The spring also acts as a low-pass filter for forces: high-frequency force spikes (like those from impacts) are absorbed and not transmitted to the environment. This is why SEAs are safer for human-robot interaction. Moreover, the spring enables force control without a dedicated force sensor. By measuring the deflection of the spring, the controller can precisely determine the force applied to the load. This is a huge advantage because force sensors can be expensive and fragile. The design of the spring stiffness (k) involves a trade-off: a stiff spring provides better bandwidth and positioning accuracy, but less compliance; a soft spring gives more compliance and lower impact forces, but reduces the actuator's ability to track fast motions. This trade-off must be balanced based on the application. Additionally, the motor-gear combination and spring placement matter. The spring can be placed after the gearbox (output side) or before it, each with different benefits. Also, the choice of spring type (helical, torsion, etc.) influences the design. In summary, the SEA converts a rigid actuator into a compliant force source, enabling safe and precise physical interaction. This principle underlies many modern collaborative robots and prosthetics.