Follow, listen, feel and go: alternative guidance systems for a walking assistance device

Federico Moro, Daniele Fontanelli, Roberto Passerone, Domenico Prattichizzo, Luca Rizzon, Stefano Scheggi, Stefano Targher, Antonella De Angeli, Luigi PalopoliarXiv 2016 · OpenReview

TL;DR

This paper introduces and evaluates four distinct guidance systems for a robotic walking assistant (c-Walker) designed to help older adults navigate safely. The systems range from active mechanical steering to passive haptic and acoustic cues, with the mechanical system proving most precise in user studies, though haptic and acoustic methods offer greater user autonomy.

What Problem Does This Paper Address?

As people age, reduced mobility can lead to a decline in confidence for independent navigation, increasing the risk of falls and social isolation. Robotic walking assistants offer a potential solution by compensating for sensory and cognitive impairments. However, a key challenge lies in how these devices effectively guide users through complex environments without being overly intrusive or compromising user autonomy. This paper tackles this by exploring and comparing different guidance modalities – mechanical, haptic, and acoustic – to find the optimal balance between precise navigation and user experience.

Methods

The researchers developed a robotic walking assistant, the "c-Walker," equipped with various sensors (encoders, IMU, cameras) and actuators. They implemented four guidance systems:

  • Mechanical Steering: The c-Walker actively steers the front wheels to guide the user. This is an "active" system where the device takes direct control of the direction.
  • Haptic Guidance: Wearable bracelets on the user's wrists provide vibrotactile feedback to suggest a turn direction. This is a "passive" system, relying on the user to interpret and act on the cues.
  • Acoustic Guidance (Left/Right): Headphones provide simple left/right audio cues to indicate direction.
  • Acoustic Guidance (Binaural): Headphones use more sophisticated binaural audio cues to create a virtual sound source, giving a more nuanced indication of direction.

The core of the guidance relies on estimating the c-Walker's position relative to a planned path, defining a "virtual corridor" rather than a strict trajectory to allow for user error and comfort. Specialized algorithms were developed for each guidance mechanism, and a within-subjects study with young participants was conducted to evaluate their performance and user experience.

Datasets

Not discussed in the paper. The study used a simulated environment with predefined virtual paths (straight, C-shaped, S-shaped) for evaluation.

Key Insights and Ideas

  • Active vs. Passive Guidance Trade-offs: The mechanical (active) system offers the highest precision but can feel intrusive. Passive systems (haptic, acoustic) provide more user autonomy but can lead to less precise navigation.
  • The "Virtual Corridor" Concept: Guiding users within a defined corridor, rather than a precise line, is crucial for user comfort and perceived control, acknowledging that perfect path following might be too restrictive.
  • Haptic and Acoustic Modalities: While both are passive, haptic feedback was generally preferred for its discreetness and acceptability in public spaces, whereas acoustic feedback (especially binaural) offered richer directional information but raised concerns about missing environmental sounds and user attention.
  • User Experience Matters: Beyond raw performance metrics, user perception of ease of use, confidence, and acceptability significantly influences the choice of guidance system.

Key Results

  • Performance: The mechanical guidance system demonstrated significantly lower average error (18.1 cm) compared to haptic (74.0 cm), acoustic (39.6 cm), and binaural (62.4 cm) systems. The straight path (I) was consistently the easiest across most guidance types.
  • Time: The mechanical system was also the fastest, with an average time of 25.4 seconds, compared to haptic (29.0 s), acoustic (26.9 s), and binaural (26.6 s).
  • Length: While most participants stayed within the target path length, the haptic system showed the largest deviation, particularly on the S-shaped path.
  • User Feedback: Participants found the mechanical system easiest to use and felt most confident with it, but also expressed concerns about it being coercive. Haptic guidance was rated as more acceptable in public spaces than acoustic/binaural systems. Headphones were generally disliked due to the potential to miss environmental sounds. Binaural guidance required more attention but was appreciated for its detailed information.

Limitations and Future Work

  • Participant Demographics: The study was conducted with young participants, not the target elderly population. Future work will involve testing with older adults to validate the methodology and gather relevant insights.
  • Ecological Validity: The experiments were conducted in a controlled environment with virtual corridors. Future research aims to repeat studies in more "ecological" contexts, with real-world environments and potential obstacles.
  • Haptic/Acoustic Algorithm Refinement: While the mechanical system excelled in performance, the passive systems showed potential for improvement. Suggestions for future work include modulating haptic vibration frequency and exploring richer acoustic cues (e.g., volume for magnitude, verbal feedback) to enhance communication.
  • Mechanical System Concerns: The paper acknowledges concerns about the mechanical system being perceived as coercive and potentially risky due to planning errors, suggesting further investigation into user acceptance and safety.
  • Alternative Actuation: The authors suggest exploring electromechanical brakes as another "active" guidance mechanism.

Summary updated

Abstract

In this paper, we propose several solutions to guide an older adult along a safe path using a robotic walking assistant (the c-Walker). We consider four different possibilities to execute the task. One of them is mechanical, with the c-Walker playing an active role in setting the course. The other ones are based on tactile or acoustic stimuli, and suggest a direction of motion that the user is supposed to take on her own will. We describe the technological basis for the hardware components implementing the different solutions, and show specialized path following algorithms for each of them. The paper reports an extensive user validation activity with a quantitative and qualitative analysis of the different solutions. In this work, we test our system just with young participants to establish a safer methodology that will be used in future studies with older adults.
BibTeX
@misc{moro2016follow,
  author = {Federico Moro and Daniele Fontanelli and Roberto Passerone and Domenico Prattichizzo and Luca Rizzon and Stefano Scheggi and Stefano Targher and Antonella De Angeli and Luigi Palopoli},
  title = {{Follow, listen, feel and go: alternative guidance systems for a walking assistance device}},
  year = {2016},
  eprint = {1601.03915},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1601.03915},
}
Figure