Dynamic Stride Length Adaptation According to Utility And Personal Space

Pedestrians adjust both speed and stride length when they navigate difficult situations such as tight corners or dense crowds. They try to avoid collisions and to preserve their personal space. State-of-the-art pedestrian motion models automatically reduce speed in dense crowds simply because there is no space where the pedestrians could go. The stride length and its correct adaptation, however, are rarely considered. This leads to artefacts that impact macroscopic observation parameters such as densities in front of bottlenecks and, through this, flow. Hence modelling stride adaptation is important to increase the predictive power of pedestrian models. To achieve this we reformulate the problem as an optimisation problem on a disk around the pedestrian. Each pedestrian seeks the position that is most attractive in a sense of balanced goals between the search for targets, the need for individual space and the need to keep a distance from obstacles. The need for space is modelled according to findings from psychology defining zones around a person that, when invaded, cause unease. The result is a fully automatic adjustment that allows calibration through meaningful social parameters and that gives visually natural results with an excellent fit to measured experimental data.

💡 Research Summary

The paper addresses a long‑standing gap in pedestrian simulation: the lack of explicit stride‑length adaptation. While most microscopic models either treat pedestrians as continuously moving particles (social‑force, differential‑equation based) or as discrete hops on a lattice (cellular automata), they typically adjust only walking speed in dense or constrained situations. Empirical observations, however, show that humans simultaneously reduce speed and shorten their steps when navigating tight corners, dense crowds, or approaching bottlenecks. Moreover, pedestrians strive to preserve a personal space around themselves, a phenomenon well documented in psychology (Hall’s model of intimate, personal, social, and public zones). Ignoring stride adaptation and personal‑space considerations leads to artefacts such as unrealistic density spikes in front of bottlenecks and reduced flow rates.

To overcome these limitations, the authors extend the Optimal Steps Model (OSM), originally introduced in their earlier work, by allowing the next foot placement to be chosen anywhere inside a disc rather than on a fixed‑radius circle. The disc radius r_i corresponds to the pedestrian’s free‑flow stride length, which is linearly related to the individual’s free‑flow speed (a relationship supported by experimental data). Within this disc, a utility function is defined that balances three competing objectives:

1. Target orientation – the desire to move toward the destination along the shortest travel time path. This is computed by solving an eikonal equation F(x)·|∇Φ(x)| = 1 with the Fast‑Marching Method, yielding an arrival‑time field Φ(x). Maximizing –Φ(x) drives pedestrians toward the target.

2. Obstacle avoidance – a static scalar field that penalizes positions inside or too close to obstacles.

3. Personal‑space preservation – a dynamic field that penalizes encroachment into the intimate (≤ 0.45 m) and personal (0.45–1.20 m) zones of other pedestrians. The authors adopt Hall’s four‑zone model but implement only the intimate and personal zones because the social and public zones have negligible influence on immediate collision avoidance. Each zone is represented by smooth compact‑support functions (exponential forms) that increase sharply as the distance to another pedestrian falls below the respective thresholds. Parameters µ_p (overall strength) and a_p (relative weight between intimate and personal zones) are introduced, giving the model socially meaningful calibration knobs.

The total utility for a candidate position x is a weighted sum: U(x) = w_target·(−Φ(x)) + w_obstacle·O(x) + w_ped·P(x), where O(x) and P(x) are the obstacle‑ and pedestrian‑avoidance fields, respectively, and the weights can be tuned per scenario.

Finding the optimal next step reduces to a two‑dimensional optimization problem over the disc. The authors employ the Nelder‑Mead simplex algorithm, which does not require gradient information and converges quickly for the smooth utility landscape. To keep computational costs low, the disc is discretized into a modest number of candidate points, and the algorithm is initialized with the previous step direction, ensuring near‑real‑time performance even for simulations with thousands of agents.

The paper validates the enhanced OSM through a series of experiments:

• Straight bottleneck – Pedestrians approach a narrow exit. The new model reproduces the experimentally observed density buildup upstream of the bottleneck without the unrealistic “jam” that occurs when stride length is fixed. Flow rates match measured values within a few percent.

• Sharp corner navigation – Agents negotiate a 90° turn. Stride shortening allows smoother turning and prevents agents from getting stuck, a problem reported in earlier OSM implementations.

• Cross‑flow and merging – The model captures realistic lane formation and density distribution when two streams intersect.

Calibration against empirical fundamental diagrams (density–speed relationships) shows that by adjusting µ_p and a_p, the model can reproduce a wide range of crowd behaviours, reflecting cultural or situational differences in personal‑space preferences.

The authors discuss computational aspects, noting that the optimization overhead scales linearly with the number of agents and remains acceptable for real‑time applications. They also acknowledge limitations: the current implementation is two‑dimensional, assumes circular agents (radius 0.20 m), and treats personal‑space radii as static. Future work could extend the framework to three dimensions (stairs, ramps), incorporate dynamic personal‑space modulation (e.g., stress‑induced contraction), and apply the model to human‑robot interaction scenarios where robots must respect human personal zones.

In conclusion, the paper presents a principled, psychologically grounded method for stride‑length adaptation in pedestrian dynamics. By formulating movement as a utility‑maximization problem on a disc and embedding Hall’s personal‑space zones into the utility, the authors achieve more realistic microscopic behaviour, which in turn yields improved macroscopic predictions of flow and density. The approach is flexible, computationally tractable, and opens avenues for richer crowd‑simulation models that bridge the gap between biomechanics, psychology, and transportation engineering.