ForceSight: Non-Contact Force Sensing with Laser Speckle Imaging

Title of the Paper

ForceSight: Non-Contact Force Sensing with Laser Speckle Imaging

Paper Information

• Subject Area: Non-contact force sensing technology, combining laser speckle imaging for human-computer interaction and ubiquitous computing
• Keywords: Force sensing, non-contact sensing, laser speckle imaging, ubiquitous computing, human-computer interaction, laser sensors, material recognition, remote operation, haptic interaction, surface deformation

Research Background and Problem

• Identified Problems or Challenges: Traditional force sensing methods rely on contact-based sensors (e.g., pressure-sensitive materials, force-sensitive resistors), which require wiring, lack flexibility, or depend on battery power, increasing costs and maintenance complexity. Additionally, these sensors are sensitive to external factors (e.g., environmental exposure), prone to inaccuracies, and require regular calibration. In scenarios involving low-cost passive objects (e.g., walls, tables) or high user experience demands (e.g., human interaction), contact-based sensors are unsuitable.

• Significance: Force is a ubiquitous signal during object contact, carrying rich information such as interactive touch actions and user behaviors. Force sensing can enhance machines' perception of users and the environment, significantly improving system robustness, accuracy, and privacy as part of multimodal sensing.

• Research Motivation and Related Work: Laser speckle technology has been used for various sensing tasks (e.g., cursor tracking, material recognition) and in medical fields (e.g., blood flow detection). Existing research primarily focuses on contact-based sensing and traditional optical methods for detecting surface deformation, with no prior work extending laser speckle technology to broad non-contact force sensing applications.

Solution

• Proposed Solution: The authors propose a non-contact force sensing method based on laser speckle imaging, "ForceSight," which identifies the magnitude of applied force by observing subtle surface deformations and the spatial and temporal trajectory changes of laser speckles.

• Innovations:

  1. Introduced a theoretical model based on laser speckle motion to describe surface deformation caused by force and the corresponding changes in laser speckles.
  2. Developed an end-to-end non-contact force sensing system, from hardware to signal processing.
  3. Designed two sensing configurations (divergent laser and focused laser) to adapt to different use cases.
  4. Systematically evaluated its performance across various common materials and distance scenarios.

• Implementation Steps:

  1. Use optical flow algorithms to estimate the Speckle Velocity Fields.
  2. Calculate the estimated applied load force for each frame through weighted aggregation and temporal integration.
  3. Derive real-time applied force from the optical flow velocity field using a regression model.

• Key Technologies:

  • Laser speckle imaging (interference and scattering of coherent light fields)
  • Optical flow methods for detecting and tracking speckle motion
  • Spatial calibration models based on material and force characteristics

Research Outcomes

• Specific Results:

  1. Validated the theoretical model for force sensing, demonstrating a linear relationship between speckle motion changes and applied force magnitude.
  2. Provided extensive experimental evaluations, covering different materials (e.g., wood, metal, acrylic), varying thicknesses, and installation positions.
  3. Developed multiple practical demonstration applications, including object recognition, non-contact touch sensing, 3D device interaction, and force sensing for remote robotic operation.

• Advantages Over Existing Solutions:

  1. Does not rely on contact-based sensors, offering flexible deployment and avoiding issues with wiring and battery power.
  2. Does not require significant modifications to the measured objects, such as embedding electronic components.
  3. Capable of long-range measurements (up to 8 meters) with high precision at sub-Newton levels (minimum error of 0.08N).

• Experimental or Evaluation Results:

  1. Short-Range Testing (Divergent Mode): On surfaces such as wood, acrylic, and metal, the average error was 0.18N (SD=0.11) and 0.31N (SD=0.12).
  2. Long-Range Testing (Focused Mode): On metal surfaces, tests conducted at distances of 2m, 4m, 6m, and 8m maintained an error of 0.18N (SD=0.03).
  3. Material Compatibility Study: Modeling was conducted on a wide range of materials, including wood, metal, foam, and silicone, demonstrating that speckle-based force sensing can map the elastic properties of materials.

• Limitations and Future Directions:

  1. Material Compatibility: ForceSight is not suitable for transparent, discontinuous, or overly rigid materials.
  2. Sensing Range and Resolution: Improvements are needed for sensing extremely large or small forces.
  3. Safety and Non-Invasiveness Optimization: Future work could explore the use of invisible infrared lasers to reduce user distraction and lower laser power to enhance safety.