Overview
Safety in shared mobility is not a behavioural problem to be corrected after deployment. It is an engineering problem that must be solved before vehicles enter public space. Shared electric vehicles operate in environments characterised by uncertainty, mixed traffic, variable infrastructure quality, and diverse user capability. In this context, safety cannot rely on static rules, user education alone, or post-incident enforcement.
Stride approaches safety as a foundational system property. It is designed into hardware, software, data architecture, and operational policy as a single, coherent framework. This document outlines how Stride engineers safety across the full lifecycle of a ride, from vehicle movement and rider behaviour to pedestrian interaction and system governance.
Safety as a Design Constraint
Public space tolerates little margin for error. Unlike private vehicles, shared micromobility systems operate continuously, across large fleets, with riders of varying experience levels. The probability of unsafe behaviour is not hypothetical. It is guaranteed at scale.
Stride treats safety as a non-negotiable design constraint, equivalent to structural integrity in civil engineering. Features are not layered on after fleet deployment. They are embedded in system architecture from the outset.
This approach rejects the assumption that compliance can be achieved solely through user agreements or punitive enforcement. Instead, the system is designed to make safe behaviour the default outcome and unsafe behaviour progressively harder to sustain.
Context-Aware Speed Control
Speed is the primary risk amplifier in shared mobility incidents. Traditional approaches rely on fixed geofences and blanket speed caps. These controls are necessary but insufficient. They assume static environments and fail to respond to real-world variability.
Stride vehicles use onboard sensing and local processing to classify their environment in real time. Road type, pedestrian density, proximity to crossings, shared path conditions, and environmental features are continuously assessed during operation.
Based on this assessment, vehicles dynamically adjust their maximum speed. In high-risk environments, speed is automatically reduced without rider input. This autoslow capability operates independently of network connectivity, ensuring consistent behaviour even in low-signal conditions.
The system prioritises harm reduction over rider convenience. Speed reductions are applied pre-emptively, not reactively. The objective is to prevent unsafe states from occurring rather than responding after thresholds are breached.
This dynamic approach allows Stride to apply finer-grained control than static zoning, reducing unnecessary restrictions in low-risk areas while strengthening controls where risk is elevated.
Rider Behaviour Modelling
Not all risk originates from the environment. Rider behaviour remains a critical variable. However, Stride does not treat riders as a homogeneous group or rely on simplistic violation counting.
Machine learning models analyse behavioural patterns across rides to identify indicators associated with increased risk. These include erratic steering inputs, repeated acceleration bursts, abrupt braking, boundary probing near geofence edges, and time-of-day risk correlations.
The system builds a probabilistic risk profile rather than a binary judgement. This allows interventions to scale proportionally. Early-stage indicators trigger in-ride prompts or enforced slow modes. Persistent or escalating patterns result in temporary restrictions or suspension.
Enforcement is automated, consistent, and transparent. It does not depend on manual reporting or discretionary intervention. This reduces bias, improves predictability, and removes pressure from councils to act as de facto enforcement agencies.
Importantly, behavioural models are continuously evaluated for false positives and drift. Safety systems that over-penalise erode trust and compliance. Model governance is therefore integral to system safety.
Impairment Risk Mitigation
Impairment remains a complex challenge in shared mobility. Stride does not pursue invasive biometric surveillance or speculative detection claims. Instead, the system focuses on observable behavioural indicators that correlate with impaired riding.
Cognitive load, reaction time variability, and control consistency are assessed within defined tolerances. When impairment risk is inferred, the system prioritises immediate harm reduction. Speed is reduced, alerts are issued, and ride termination may occur if risk thresholds are exceeded.
These mechanisms are designed to be preventative rather than punitive. The objective is to interrupt unsafe riding before incidents occur, not to accumulate evidence after harm has been done.
All impairment-related systems are tested under controlled conditions and evaluated in collaboration with research institutions. Deployment is gated by demonstrated efficacy and bias assessment, not marketing pressure.
Pedestrian Safety as a Primary Requirement
Pedestrians are the most vulnerable participants in shared mobility environments. Safety systems that focus exclusively on riders fail to address the asymmetric risk borne by people on foot, particularly those with disability.
Electric vehicles generate minimal acoustic cues at low speed, significantly reducing detectability. This creates heightened risk for blind and low-vision pedestrians.
Stride vehicles incorporate low-speed audible alert systems designed to address this gap. These alerts are activated during movement initiation and low-speed travel, providing early warning without excessive noise pollution. Sound profiles are engineered for detectability rather than volume.
Visual signalling supplements audio alerts through LED systems configured for visibility in low-light and high-traffic conditions. These signals are designed to communicate presence, not intent, reducing ambiguity for pedestrians.
Stride’s pedestrian safety features are informed by direct collaboration with accessibility organisations, including the Canberra Blind Society. Feedback from affected users is integrated into design iteration and evaluation.
Pedestrian safety is treated as a first-order requirement, not an externality of rider behaviour.
Rider Awareness and In-Ride Guidance
Safety systems are only effective if riders understand and respond to them. Stride’s mobile applications function as real-time safety interfaces rather than passive booking tools.
Riders receive context-aware alerts when entering zones requiring reduced speed or heightened caution. These alerts are delivered using audio and visual cues optimised for comprehension while in motion.
The interface prioritises brevity and clarity. Safety prompts are designed to interrupt behaviour without overwhelming the rider. Persistent cues reinforce compliance without habituation.
This approach reduces reliance on pre-ride tutorials or static signage, which are poorly retained under real-world conditions.
Vehicle Hardware Integrity
Software safety controls are ineffective without reliable hardware. Stride’s safety framework includes strict vehicle maintenance and monitoring standards.
Vehicles continuously report system health metrics, including braking performance, battery status, sensor functionality, and structural integrity indicators. Anomalies trigger automated removal from service pending inspection.
Maintenance cycles are informed by usage intensity rather than fixed schedules. High-stress vehicles receive earlier intervention, reducing failure risk under peak conditions.
This predictive maintenance approach reduces the likelihood of mechanical failure contributing to safety incidents.
System Observability and Auditability
Automated safety systems operating in public space must be accountable. Stride’s platform logs all safety-relevant decisions with contextual metadata.
Speed reductions, behavioural interventions, enforcement actions, and system overrides are recorded in a structured, reviewable format. This enables post-incident analysis, model refinement, and external audit.
Authorised government and council officials have access to aggregated safety metrics through secure dashboards. These include compliance rates, incident correlations, intervention frequency, and spatial risk patterns.
This level of observability supports evidence-based policy decisions and continuous system improvement. It also provides transparency necessary for public trust.
Avoiding Black-Box Enforcement
Black-box enforcement undermines legitimacy. Stride explicitly avoids opaque decision-making that cannot be explained or reviewed. Safety interventions are governed by documented thresholds and model logic. Where machine learning is used, outputs are interpretable and subject to override if systemic issues are identified.
Appeal and review mechanisms are built into operational workflows. Riders are not subject to irreversible penalties without review pathways. This approach recognises that automated systems must remain subordinate to accountable governance, particularly when operating in shared civic environments.
Safety and Urban Integration
Micromobility systems do not operate in isolation. Their safety performance is inseparable from urban design, infrastructure quality, and regulatory frameworks. Stride’s safety platform is designed to integrate with city planning processes. Geospatial analytics identify high-risk corridors, recurring conflict zones, and infrastructure gaps.
These insights support targeted interventions such as speed zone adjustments, parking redesign, or infrastructure investment. Safety data is not merely reported. It is actionable.
By aligning system behaviour with city objectives, Stride reduces friction between operators and regulators.
Conclusion
Safety in shared mobility cannot be achieved through isolated features or reactive enforcement. It requires systems engineered for uncertainty, scale, and public accountability.
Stride’s approach treats safety as infrastructure. Speed is context-aware. Behaviour is modelled, not assumed. Pedestrians are prioritised. Enforcement is automated but auditable. Hardware and software are governed as a single system.
