Easing the Path: Predicting Pedestrian Comfort Around Robots

Author: Denis Avetisyan

New research quantifies how robot movement impacts pedestrian comfort levels, paving the way for more natural and safer human-robot interactions.

The study defines key spatial kinematic variables-minimum relative Euclidean distance [latex]D_{min}[/latex], lateral distance at passing [latex]D_{lat}[/latex], maximum trajectory curvature ρ, and distance at the point of temporal proximity [latex]D_{T_{P}}[/latex]-to characterize the robot’s navigation around pedestrians, acknowledging that each parameter contributes to a predictable geometry of potential failure in dynamic human-robot interaction.

This study identifies projected time-to-collision as a crucial kinematic variable for accurately predicting pedestrian comfort and proposes a composite predictor for improved robot path planning.

While mobile robots increasingly share public spaces with pedestrians, ensuring comfortable interactions remains a significant challenge beyond simply avoiding collisions. This is addressed in ‘Empirical Prediction of Pedestrian Comfort in Mobile Robot Pedestrian Encounters’, which investigates the relationship between robot kinematics and subjective human comfort through one-on-one experimental trials. The study demonstrates a statistically significant correlation between variables like minimum distance and projected time-to-collision (PTTC) with pedestrian comfort, culminating in a composite predictor with an odds ratio of 3.67. Could these findings pave the way for more socially aware path planning and truly comfortable human-robot coexistence?

The Illusion of Safety: Navigating Human Spatial Expectations

The increasing presence of social robots in human-populated spaces necessitates a shift in design priorities, moving beyond simply avoiding collisions to actively fostering pedestrian comfort. As these robots transition from controlled environments to shared public areas – sidewalks, shopping malls, and even homes – their acceptance hinges not only on functional safety, but on how comfortably humans perceive their presence and movements. This comfort is not merely a subjective feeling; it directly impacts a person’s willingness to interact with the robot, cooperate with its tasks, and ultimately, integrate it into daily life. Therefore, understanding the factors that contribute to a feeling of spatial security and psychological ease around robots is paramount for successful human-robot collaboration and widespread adoption.

While robotic safety has long been a primary design consideration, research indicates that ensuring pedestrian comfort is equally vital for successful human-robot interaction. Traditional metrics, such as collision avoidance and minimum stopping distances, establish a baseline for physical safety, but fail to account for the psychological factors influencing how people feel around robots. A robot adhering strictly to safety protocols may still be perceived as unsettling if it approaches too closely, moves unpredictably, or fails to respect personal space. This perceived discomfort can lead to avoidance behaviors, diminished trust, and ultimately, hinder the robot’s ability to function effectively in shared environments. Consequently, designs must integrate an understanding of proxemics, kinesics, and social norms to create robots that are not only safe, but also welcomed and readily accepted by pedestrians.

Current methodologies for assessing robot-pedestrian interaction frequently treat robot movement as purely mechanical, neglecting how humans perceive and interpret those motions within their personal space. This simplification overlooks critical aspects of human spatial cognition; individuals don’t react solely to a robot’s physical proximity or velocity, but also to its implied trajectory and the perceived effort behind its maneuvers. Research indicates that jerky, unpredictable, or overly deliberate robot kinematics can induce discomfort even if a collision is physically impossible. The human brain constantly predicts future positions based on observed movement patterns, and deviations from these predictions – particularly those suggesting a lack of intention or control – trigger a negative affective response. Therefore, a more holistic approach is needed, one that integrates principles of human spatial perception and anticipates how subtle variations in robot motion will be interpreted by pedestrians.

The successful integration of robots into everyday human environments hinges not only on their ability to avoid collisions, but also on fostering a sense of comfort among pedestrians. Currently, robot designs largely prioritize safety – ensuring physical avoidance of obstacles – yet fail to adequately address the perceived comfort of those sharing the space. Establishing reliable predictors of pedestrian comfort – encompassing factors like proxemics, movement predictability, and even aesthetic design – is therefore essential. These predictors would move beyond simple collision avoidance to inform robot behaviors that are not merely safe, but also intuitively acceptable and even welcoming, ultimately determining whether these machines are seen as helpful collaborators or intrusive obstacles. Without this focus, even flawlessly safe robots risk being rejected by the very people they are intended to serve.

Subjective comfort scores exhibit correlations with robot speed, minimum approach distance, lateral distance, maximum curvature, and metrics related to the patient-tool contact time [latex]T_{P}[/latex], including both the minimum contact time and the distance at which it occurs.

Kinematic Signatures: The Language of Robotic Comfort

Robot speed and minimum approach distance are primary determinants of perceived safety and comfort for individuals encountering a robotic system. Studies indicate a strong inverse relationship between robot velocity and acceptable proximity; higher speeds necessitate greater distances to avoid eliciting avoidance behaviors or feelings of threat. Specifically, maintaining a minimum distance of at least 0.5 meters is generally considered crucial for establishing a comfortable personal space bubble, though this threshold is affected by the robot’s speed and the environment. Failing to adhere to these principles can result in increased anxiety, defensive maneuvers by pedestrians, and a negative perception of the robot’s intent, even if the robot’s path is technically collision-free. These factors are essential considerations in the design of socially aware robotic navigation systems.

Lateral distance, the robot’s displacement perpendicular to the pedestrian’s primary travel direction, significantly impacts perceived comfort and acceptance. Studies indicate that pedestrians react negatively to robots exhibiting large or unpredictable lateral movements, even when the overall approach velocity and minimum distance are maintained at safe levels. This discomfort arises from the violation of expected spatial boundaries and introduces uncertainty regarding the robot’s intent. Specifically, consistent deviations from a linear path, or rapid changes in lateral position, elicit stronger avoidance behaviors and increased physiological responses indicative of stress in observers. Minimizing lateral displacement and ensuring smooth, predictable side-to-side movement are therefore critical for fostering positive human-robot interactions and maximizing pedestrian comfort.

Trajectory curvature, as a kinematic variable, directly impacts human perception of a robot’s approach and intent. Higher curvature, indicating sharper turns or more erratic movement, correlates with reduced perceived comfort and predictability. Studies demonstrate that pedestrians anticipate robot paths based on initial trajectory segments; abrupt changes in curvature necessitate increased cognitive load as individuals recalculate potential collision risks. A smoother, lower-curvature path allows for more accurate prediction of the robot’s future position, fostering a sense of safety and enabling pedestrians to react appropriately. Quantitatively, curvature can be expressed as the rate of change of the tangent angle along the path, with lower values indicating a straighter, more predictable trajectory.

The relationship between a robot’s kinematic variables – speed, distance, and trajectory – and human comfort is theoretically grounded in spatial perception research. Hall’s Proxemics Theory posits that humans maintain comfortable distances during interactions, categorized into intimate, personal, social, and public zones; robot behavior that violates these perceived boundaries can induce discomfort. This is computationally modeled through frameworks like the Social Force Model (SFM), which treats pedestrians as particles affected by attractive and repulsive forces. Within SFM, robot kinematics directly influence these force calculations; for example, a robot’s speed impacts the magnitude of the repulsive force experienced by a pedestrian, while trajectory curvature affects the predictability of the robot’s movement and thus a pedestrian’s ability to anticipate and react. Consequently, optimization of robot motion must account for these spatially-derived perceptual thresholds to ensure a comfortable and safe human-robot interaction.

Utilizing an Agilex Scout Mini base with [latex]Zedx[/latex] depth and 3D LiDAR sensors, the robot successfully navigates head-on approaches and passes pedestrians as demonstrated in the supplementary video.

Predictive Algorithms: Modeling the Illusion of Intent

Three distinct algorithms were implemented to predict pedestrian comfort levels in proximity to the robot. The Minimum Distance Predictor calculated comfort based solely on the shortest distance between the robot and a pedestrian. The PTTC-Based Estimator (Predicted Time To Closest Approach) assessed comfort using the calculated time until the minimum distance was reached, providing a dynamic measure of potential collision risk. Finally, the Composite Comfort Predictor integrated data from both of these methods, alongside additional kinematic variables, to provide a more comprehensive assessment of the interaction.

The Composite Comfort Predictor utilizes a multi-variable approach to assess human-robot interaction comfort levels. This predictor incorporates robot speed, the minimum distance between the robot and a pedestrian, the magnitude of lateral robot movement, trajectory curvature as a measure of path predictability, and the minimum Predicted Time To Collision (PTTC) – the shortest duration before a potential collision. By integrating these kinematic variables, the predictor aims to provide a more comprehensive evaluation of comfort than single-variable methods, accounting for both proximity and the dynamic characteristics of the robot’s movement relative to the pedestrian.

Data acquisition for this study utilized an Agilex Scout Mini robot platform equipped with a 3D LiDAR and a Zedx Depth Camera. The LiDAR provided precise environmental mapping and distance measurements, while the Zedx Depth Camera facilitated detailed perception of both the robot’s and pedestrians’ positions in three-dimensional space. This sensor suite enabled the accurate recording of kinematic data, including velocity, trajectory, and proximity, with a reported positional accuracy of ±2 centimeters. Raw sensor data was then processed to extract relevant features for comfort prediction, specifically focusing on the relative movements between the robot and observed pedestrians within the testing environment.

The Minimum Predicted Time To Collision (PTTC) was calculated continuously throughout robot-pedestrian interactions to quantify the closest spatial and temporal proximity achieved between the two. This metric represents the time remaining until a collision would occur if both parties maintained their current velocities and trajectories. A lower PTTC value indicates a more critical approach, directly correlating with increased perceived risk for the pedestrian and a corresponding decrease in comfort level. The PTTC was calculated using positional data acquired from 3D LiDAR and depth cameras, allowing for precise determination of inter-agent distance and relative velocity, and was weighted within the Composite Comfort Predictor alongside other kinematic variables to provide a comprehensive comfort assessment.

Comfort scores demonstrate a clear correlation with the magnitude of the middle variable E.

Statistical Validation: Discerning Signal from Noise

The Chi-Square Test was employed to determine the statistical significance of the relationship between each predictor variable and accurately estimated pedestrian comfort levels. This test assesses whether observed frequencies of comfortable versus uncomfortable experiences differ significantly from what would be expected by chance, given the predictor values. Specifically, contingency tables were constructed for each predictor, comparing predicted comfort levels based on the model with the actual observed comfort levels. The resulting Chi-Square statistic and associated p-value indicate the probability of observing the obtained results if no association existed between the predictor and comfort. A p-value below the significance threshold (typically 0.05) suggests a statistically significant association, indicating the predictor reliably contributes to estimating comfort.

The Odds Ratio, a measure of association between a predictor variable and pedestrian comfort levels, was calculated for each tested variable. This ratio quantifies the relative likelihood of observing a specific comfort level given the presence or absence of a particular predictor. An Odds Ratio of 1 indicates no association, values greater than 1 suggest a positive correlation – increasing the odds of a comfortable experience – and values less than 1 indicate a negative correlation. The magnitude of the ratio reflects the strength of this association; a higher absolute value indicates a stronger predictive relationship between the variable and observed comfort.

The F1 Score was employed as a composite metric to evaluate predictive model accuracy by considering both precision and recall. Precision measures the proportion of correctly predicted comfortable pedestrian experiences out of all instances predicted as comfortable, while recall represents the proportion of actual comfortable experiences that were correctly identified. The F1 Score is calculated as the harmonic mean of precision and recall, providing a balanced assessment of the model’s ability to avoid both false positives and false negatives. A higher F1 Score indicates better overall performance, particularly in scenarios where class imbalance exists – a common characteristic of datasets analyzing subjective experiences like pedestrian comfort. In this analysis, the Composite Comfort Predictor achieved an F1 Score of 0.72, demonstrating a robust balance between precision and recall in identifying comfortable pedestrian experiences.

Statistical analysis demonstrates the Composite Comfort Predictor’s superior performance in estimating pedestrian comfort levels. The model achieved an Odds Ratio of 3.67, signifying a 3.67-fold increase in the likelihood of correctly identifying comfortable experiences compared to other models tested. Furthermore, the Composite Predictor’s F1 Score of 0.72 indicates a balanced combination of precision and recall in its predictions, confirming robust predictive accuracy. The Chi-Square test, yielding a p-value less than 0.0006, establishes the statistical significance of these results, supporting the conclusion that the Composite Comfort Predictor is a reliable tool for assessing pedestrian comfort.

Toward Empathetic Machines: A Future of Seamless Coexistence

The successful integration of robots into human environments hinges not simply on avoiding collisions, but on fostering a sense of comfort and predictability for those sharing the space. Recent investigations highlight that a robot’s kinematic factors – encompassing its speed, acceleration, turning radius, and overall movement patterns – exert a surprisingly strong influence on how humans perceive its behavior. Researchers are discovering that even subtle variations in these parameters can dramatically alter a pedestrian’s sense of personal space and anticipatory ability. A comprehensive understanding of these kinematic influences is therefore crucial; robot designs that prioritize smooth, predictable motions, and account for the nuances of human spatial awareness, are far more likely to be accepted and trusted in shared environments, ultimately leading to more harmonious human-robot interactions.

Ongoing research aims to refine comfort prediction models for robots operating in shared spaces by moving beyond generalized assessments and embracing the nuances of individual pedestrian preferences. This involves incorporating data relating to personal space expectations, walking speed, and even cultural backgrounds, recognizing that comfort is subjective and varies significantly between individuals. Furthermore, the models will integrate contextual factors such as the environment – crowded sidewalks versus open plazas – and the robot’s purpose, allowing for a more accurate anticipation of how a pedestrian might perceive its approach. By accounting for these multifaceted influences, future robots will not simply avoid collisions, but proactively adjust their behavior to create a genuinely comfortable and positive experience for everyone they encounter.

The potential for robots to dynamically modify their movements based on immediate human feedback represents a crucial step towards seamless human-robot interaction. Rather than adhering to pre-programmed paths, these robots could utilize real-time comfort assessments – gauging proxemic preferences and predicting potential discomfort – to adjust both speed and trajectory. This adaptive behavior extends beyond mere safety; it aims to create a more intuitive and agreeable experience for pedestrians. By continuously refining its movements in response to subtle cues, a robot can minimize feelings of intrusion or anxiety, fostering a sense of shared space rather than one of mechanical imposition. Ultimately, this responsiveness could transform robots from functional tools into genuinely considerate companions, significantly enhancing the quality of interactions and encouraging broader acceptance of robotic presence in everyday life.

The successful integration of robots into everyday human environments hinges not merely on their ability to navigate safely and perform tasks efficiently, but also on fostering a sense of comfort and trust amongst those sharing those spaces. Prioritizing pedestrian comfort – considering factors like personal space, predictable motion, and perceived politeness – moves beyond simply avoiding collisions and towards a more nuanced understanding of human-robot interaction. This approach suggests that robots designed with a focus on minimizing discomfort and maximizing positive social signals are more likely to be accepted and welcomed, transitioning from perceived as tools to being viewed as cooperative partners. Ultimately, cultivating this sense of ease and trust is crucial for unlocking the full potential of social robotics and ensuring these technologies genuinely enhance the human experience.

The study’s focus on predicting pedestrian comfort through kinematic variables echoes a fundamental truth about complex systems. It isn’t about achieving a static ‘comfortable’ state, but rather anticipating the inevitable decay of any initial arrangement. As Henri Poincaré observed, “It is through science that we arrive at certainty, and through art that we arrive at probability.” This research doesn’t seek to build comfortable interactions; it seeks to probabilistically forecast them, acknowledging that projected time-to-collision, as a key variable, is only a snapshot in a constantly shifting dynamic. Belief in a perfect comfort threshold is a denial of entropy; the system will always tend toward disorder, and adaptation – prediction – is the only viable response.

What’s Next?

The pursuit of predictable pedestrian comfort, as demonstrated by this work, reveals a deeper truth: a perfectly comfortable pedestrian is a pedestrian who has never truly encountered a robot. The composite predictor, while a refinement, merely formalizes the art of anticipating human reaction-a system destined to be perpetually out of sync with the inherent unpredictability of organic life. Each optimized path, each minimized discomfort score, is a temporary reprieve, delaying the inevitable moment when a novel interaction exposes the model’s limitations.

Future iterations will undoubtedly focus on incorporating more nuanced behavioral models, perhaps drawing on principles of social psychology or even attempting to predict individual variations in comfort thresholds. However, such efforts are akin to building a more detailed map of a shifting landscape. The terrain will always change. The true challenge lies not in predicting reaction, but in designing systems that respond gracefully to unexpected behaviors – systems that embrace the possibility of discomfort as a signal for adaptation.

A system that never breaks is, after all, a dead system. The goal should not be to eliminate discomfort, but to cultivate a robust ecosystem where even moments of friction contribute to a more resilient and ultimately, more human-compatible robotic presence. Perfection, in this context, leaves no room for people.

Original article: https://arxiv.org/pdf/2604.13677.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/