TL;DR
This paper introduces a novel distributed pressure sensing system for lower-limb exoskeletons, designed to provide a more comprehensive understanding of the human-robot interface than traditional single-point force measurements. By integrating an array of soft, opto-electronic sensors, the system offers a detailed map of interaction pressure, crucial for enhancing safety, comfort, and overall effectiveness in rehabilitation robotics.
What Problem Does This Paper Address?
Exoskeletons, particularly those used in rehabilitation, rely on close physical coupling with the user. Quantifying this human-robot interaction is vital for ensuring safety and comfort, especially where high pressures can lead to discomfort or pain. Current methods primarily focus on measuring interaction forces or estimating interaction torques. However, these approaches often fail to capture the distribution of pressure across the entire contact area between the user's limb and the exoskeleton. This lack of detailed pressure information can mask localized high-pressure points, which are critical for user comfort and can impact the effectiveness of rehabilitation. Furthermore, some existing methods struggle with distributed contact areas or can be affected by the complex dynamics of soft tissues and compliant interfaces. This paper aims to bridge this gap by developing a system that directly measures distributed pressure at the interface.
Methods
The core of the proposed solution is a novel soft tactile sensor based on a mechano-opto-electronic transduction principle, an adaptation of the lab's "Skilsens" technology. Each sensor element consists of a light emitter and receiver. When pressure is applied, the sensor deforms, occluding the light path and reducing the light reaching the receiver, which changes its output current. These elements are arranged in an 8-channel array, housed within a flexible silicone shell.
The researchers designed and fabricated an 8-channel "Skilsens pad" with specific geometric parameters optimized through finite element analysis (FEA) to achieve a desired force range (up to 60 N) and pressure sensitivity (around 50 kPa). The material used was a shore A 40 silicone.
For the application, these pads were integrated into a custom rigid plastic frame that attaches to the flexible belt of a lower-limb exoskeleton (LOPES). This frame positions the sensors between the user's thigh and the belt, ensuring the soft silicone is in direct contact with the skin. The system was calibrated by comparing the sensor outputs with measurements from a 6-axis load cell under both static and dynamic loading conditions. An estimation algorithm was developed to translate the voltage outputs from the eight sensor channels into force and pressure distribution values.
Datasets
Not discussed in the paper. The paper focuses on the development and characterization of a new sensor system and its application.
Key Insights and Ideas
- Distributed Pressure is Key: The paper strongly advocates for distributed pressure sensing over single-point force measurements for human-robot interfaces. This is crucial for understanding user comfort and safety, as localized high pressures can be problematic even if the overall force is within acceptable limits.
- Soft Sensors for Comfort: The use of soft silicone sensors directly in contact with the skin is a deliberate choice to maximize comfort and minimize invasiveness, a critical factor in rehabilitation devices.
- Opto-electronic Transduction: The mechano-opto-electronic principle offers a robust and potentially low-cost sensing mechanism, leveraging light occlusion for pressure detection.
- Calibration is Crucial and Subject-Specific: The study highlights that sensor performance is dependent on loading conditions and individual subject characteristics (e.g., thigh shape, belt fastening). Therefore, a personalized calibration process is essential for accurate measurements.
- Preloading is Measurable and Important: The sensors can effectively measure the "preloading" force exerted by the belt fastening. This is valuable for ensuring consistent and reliable cuff application by therapists.
- Dynamic Performance is Promising: The sensors demonstrated good performance under dynamic loading conditions, suggesting their suitability for real-time monitoring during movement tasks.
Key Results
- The developed 8-channel Skilsens pad demonstrated a force range of up to 60 N with an average stiffness of 40 N/mm, closely matching FEA predictions.
- The sensor exhibited a non-amplified gain of approximately 3.3 mV/N, with a normalized RMSE and maximum percentage error generally below 5% for individual channels.
- Static calibration showed that a constant fraction of interactive force (ranging from 7-21%) was unloaded on each pad, depending on its position and the subject. This linear relationship allowed for reliable estimation of total interactive force.
- Preloading forces varied significantly across subjects and pads, ranging from 4-18 N, highlighting the importance of this measurable parameter for consistent cuff application.
- Dynamic characterization showed that the sensors did not suffer significant attenuation effects up to 3 Hz, indicating their viability for monitoring during movement.
- In a walking task simulation, the system successfully captured pressure distribution variations throughout the gait cycle, revealing localized pressure peaks not detectable by a single load cell.
Limitations and Future Work
- Partial Coverage: The prototype only covered a fraction of the contact area. The authors acknowledge that slippage between the belt and the thigh could invalidate calibration. Future work should aim to cover the full interaction area to eliminate this issue.
- Subject Representativeness: The experiments were conducted on four healthy male subjects, who are not representative of the typical population for rehabilitation protocols. Future work should involve testing with a more diverse subject pool.
- Material Friction: The FEA model simplified the contact between the indenter and the silicone as frictionless due to the difficulty in modeling friction on hyperelastic materials.
- Future Directions: The authors suggest that this system could be extended to other exoskeleton devices and potentially used to detect user tissue deformations. They also propose exploring its use as a low-cost alternative to more invasive muscular activity measurements like surface EMG.
Summary updated