Abstract
Safety is one of the most important issues in rehabilitation robot suits. We have designed a rehabilitation robot suit equipped with two mechanical safety devices called the "Velocity-based safety device" and the "torque-based safety device". The robot suit assists a patient's knee joint. The safety devices consist of only passive mechanical components without actuators, controllers, or batteries. The torque-based safety device stops the robot suit if it detects an unexpected high joint torque. Similarly, the velocity-based safety device stops the robot suit if it detects an unexpected high joint angular velocity. The mechanism for detecting the unexpected angular velocity in the velocity-based safety device is a mass-spring-damper system. In this paper, we firstly review the rehabilitation robot suit equipped with the safety devices. Secondly, we design and develop the mechanism for detecting the unexpected angular velocity by considering the natural frequency. Finally, we present experimental results to verify the effectiveness of the velocity-based safety device with the developed mechanism.
Introduction
In the field of wearable robotics, particularly for rehabilitation, safety is the highest priority. If a suit assisting a patient's knee joint suffers a software glitch or hardware failure, it could move too fast or with excessive force, leading to injury. While most modern robots rely on software-based limits, software can crash and sensors can fail.
To address this, researchers have developed a "mechanical guardian": a pair of passive safety devices that require no electricity to function. By using a fail-safe mechanical layer—an analog backup independent of sensors, controllers, or batteries—the suit can physically cut its own power if it detects a "runaway" event (excessive speed or torque).
The Mechanical Safety Mechanism
The team focused on a Velocity-Based Safety Device integrated into the knee-assist robot suit. This device uses a mass-spring-damper system to detect dangerous speeds. A bar is connected to the robot's drive shaft via a rotary damper and held in place by two tension springs.
As the shaft turns, the rotary damper exerts a force on the bar proportional to the speed. If the speed exceeds a pre-set limit, the damping force overcomes the spring tension, causing the bar to swing and physically hit a kill switch that cuts the motor's power. To prevent "false positives," the authors designed the bar's mass and spring constants so that the device would not resonate at typical human walking frequencies (roughly 0–2 Hz).
Key Insights and Experimental Results
By ensuring the natural frequency of the safety bar (calculated at 315 rad/s) is far higher than the frequency of walking (approx. 13 rad/s), the researchers ensured the bar stays steady during normal use but reacts instantly to a spike in speed. The trigger speed is adjustable by changing the position of the limit switch or swapping out the tension springs.
Experimental validation yielded the following results:
- Reliability: In 10 out of 10 trials, the device successfully cut power when the gear reached the target "runaway" velocity of 8 rad/s.
- Stability: During a simulated one-hour walking test, the device recorded zero false triggers.
- Low Latency: Experimental data showed that the bar's movement tracked the gear's velocity with almost zero time delay at walking frequencies, confirming the theoretical frequency analysis.
Conclusion
This research presents a successful proof-of-concept for a single joint (the knee) under controlled conditions. By using only springs, dampers, and physical switches, the safety system remains active even if the robot's battery dies or the CPU freezes. Future work will focus on miniaturizing the mechanical assembly for better integration into the suit frame and expanding the system to handle multi-joint coordination in full-body suits.
Summary updated