Micromobility Safety Compliance Technology: What Fleet Operators Must Know

In September 2023, a residential building fire in New York City's Bronx neighbourhood killed four people and injured dozens more. Investigators traced the cause to a lithium-ion battery from a rental e-bike. That single incident accelerated what cities had been debating for years: micromobility vehicles are not toys, not novelties, and not adequately governed by the patchwork of rules that most markets currently apply to them. Micromobility safety compliance technology is now the frontier where hardware certification, real-time AI monitoring, and evolving regulation are colliding at speed.

The fundamental tension in this space is that micromobility has scaled faster than the governance structures built to contain it. By the end of 2022, shared e-scooter services alone had logged over 200 million trips across Europe, according to the European Commission's urban mobility data. Yet the standards applied to the vehicles making those trips were drawn up for different product categories: bicycles, mopeds, or generic consumer electronics. Manufacturers were certifying e-scooters against standards never written with them in mind, fleet operators were deploying hardware with uncertain safety margins, and cities were issuing permits without clear technical criteria. The compliance picture has improved since then, but it remains fragmented, inconsistently enforced, and increasingly urgent as battery fires, collision data, and pedestrian injury reports accumulate.

What Safety Standards Actually Apply to E-Scooters and E-Bikes Today

European e-bike regulation has the most established foundation. The EN 15194 standard, developed by the European Committee for Standardisation (CEN), covers electrically power-assisted cycles (EPACs) and specifies requirements for motor power, speed cut-off at 25 km/h, braking performance, electrical safety, and electromagnetic compatibility. Products sold in the EU market must carry CE marking demonstrating conformity with the applicable directives, including the Machinery Directive 2006/42/EC or, for newer categories, the General Product Safety Regulation (GPSR) that replaced the older General Product Safety Directive and came into force in December 2024. For e-scooters, the picture is messier. There is no single harmonised European product standard equivalent to EN 15194. CEN and CENELEC have been working on EN 17128, which covers personal light electric vehicles (PLEVs) including e-scooters, and the standard defines mechanical strength, braking distance, lighting, and electrical requirements. However, EN 17128 adoption across member states remains uneven, and some manufacturers have continued to certify products under older, less specific standards. The United Kingdom, post-Brexit, has retained much of the CE framework through UKCA marking, but trials for e-scooter use on public roads under the Road Traffic Act have created a unique domestic situation where privately owned e-scooters remain technically illegal on public highways while rental schemes operate under local authority permits. In the United States, the Consumer Product Safety Commission (CPSC) regulates e-scooters under the Consumer Product Safety Act, and the voluntary standard UL 2272 covers electrical systems for personal e-mobility devices. UL 2272 has become a de facto entry requirement for many city procurement programmes since its introduction, but it is not a federal mandate. E-bikes in the US are governed by the Consumer Product Safety Improvement Act of 2008, which established a three-class system defining low-speed electric bicycles by motor wattage and assisted speed. The practical consequence of this standard fragmentation is that a vehicle certified in one jurisdiction may not meet the specific technical thresholds required in another, creating significant compliance overhead for OEMs and fleet operators operating across borders.

It is worth being honest about what standards do and do not guarantee. Certification against EN 17128 or UL 2272 confirms that a vehicle met certain performance and electrical safety thresholds at the time of manufacture and testing. It says nothing about the vehicle's condition after six months of rental use, nothing about rider behaviour, and nothing about the interaction between that vehicle and a given urban environment. A scooter that passes a braking distance test on a dry test surface in a lab may behave quite differently on a wet cobbled street in Edinburgh or a rutted cycle lane in Lisbon. Compliance, in other words, is a starting point, not an ongoing guarantee.

Battery Safety and Li-Ion Compliance: The Regulation Gap That Kills People

Lithium-ion battery fires in micromobility vehicles have become a measurable public safety crisis in several major cities. In London, the London Fire Brigade reported attending 155 e-bike fires in 2023, up from 87 in 2021, a near-doubling in two years driven by the proliferation of cheaper battery packs of uncertain provenance. The New York City Fire Department tracked 268 fires and 36 deaths related to lithium-ion batteries in micromobility devices between 2019 and 2023. These figures are not speculative risk assessments; they are recorded fatalities. The root cause in most cases is not sophisticated: counterfeit or substandard battery cells, battery management systems (BMS) that fail to prevent thermal runaway, chargers that do not match battery specifications, or physical damage to cells that accumulates through normal rental use without adequate inspection. The relevant compliance frameworks for battery systems include IEC 62133 (the international standard for secondary lithium cells and batteries used in portable applications), UN 38.3 (which governs the transport of lithium batteries and includes a series of abuse tests including altitude simulation, thermal cycling, vibration, shock, and overcharge), and, in the European context, the EU Battery Regulation (Regulation EU 2023/1542), which came into full effect in stages from 2023 onwards. The EU Battery Regulation is the most significant piece of battery-specific legislation the sector has faced. It introduces mandatory carbon footprint declarations, due diligence requirements for raw material sourcing, minimum recycled content thresholds phased in from 2027 and 2031, and digital battery passports for industrial and EV batteries. For micromobility specifically, the regulation's requirements around safety, labelling, and end-of-life management are directly applicable to the lithium batteries installed in e-scooters and e-bikes sold or deployed in the EU. CE marking on a battery or battery-powered vehicle must now reflect conformity with this regulation's safety annexes, not merely legacy directives.

The honest qualification here is that regulation alone cannot prevent battery fires if the enforcement infrastructure is not present to back it up. The London and New York fires predominantly involved privately purchased, lower-cost vehicles or aftermarket battery replacements, not vehicles from large regulated rental fleets with formal procurement standards. The compliance frameworks that exist are reasonably well-designed for the formal supply chain; the problem is the volume of product entering markets through informal channels where enforcement is limited. For fleet operators, this creates a specific reputational and operational risk: a fire involving a vehicle in their livery, even one with genuine certification, damages public confidence across the category. Some operators have responded by adopting battery health monitoring systems that track cell voltage variance, internal resistance, and temperature across their fleets in real time, flagging packs that show early indicators of degradation before they reach thermal runaway conditions.

Certification, Labelling, and Testing: How the Process Actually Works

For a manufacturer bringing an e-scooter or e-bike to the European market, the certification pathway begins with identifying which directives and regulations apply to the product. An e-bike falling under the EPAC definition in EN 15194 typically follows a self-declaration route: the manufacturer conducts or commissions testing at an accredited third-party laboratory, compiles a technical file demonstrating conformity, and then issues a Declaration of Conformity before affixing CE marking. The technical file must be held and made available to market surveillance authorities for ten years. For products outside the EPAC definition, such as higher-powered e-bikes or speed pedelecs, different type-approval processes under the L-category vehicle framework (Regulation EU 168/2013) may apply, which involves more formal type approval through national approval authorities. EN 17128 for PLEVs follows a broadly similar self-declaration model, but because the standard is still achieving market uptake, some notified bodies offer third-party certification as an additional credibility signal, particularly for manufacturers targeting public sector contracts or city operator partnerships where procurement criteria may specify independent verification. In the United States, UL certification for e-scooters under UL 2272 is voluntary but has become commercially necessary in several major markets. UL testing covers the electrical drive train system, including motors, controllers, wiring harnesses, connectors, and the battery and charging system. Products that pass receive a UL Listing Mark, which many city permit programmes now require as a baseline condition. Testing typically involves a combination of construction reviews, electrical safety tests, mechanical abuse tests, and environmental exposure tests. The entire process from initial submission to certification can take several months and costs vary widely depending on the complexity of the product and the testing scope required. Labelling requirements are a specific compliance area that is sometimes underestimated. Under the EU Battery Regulation, batteries must carry labels indicating capacity, chemistry, and collection symbols. Under the GPSR, products must carry clear warnings, instructions for safe charging, and identification markings that allow traceability back through the supply chain. In several EU member states, e-scooters offered for rental must also carry visible identification allowing users to report specific vehicles, a requirement that has created a secondary market for durable labelling solutions rated for outdoor use across varying weather conditions.

Testing processes have their own structural limitations worth acknowledging. Crash testing and biomechanical analysis for micromobility is considerably less developed than for the automotive sector. There is no equivalent of the Euro NCAP programme providing independent, standardised, publicly reported crash safety ratings for e-scooters or e-bikes. Some research has been conducted through academic and public health routes: a 2021 study published in the journal Injury Prevention analysed 116,000 e-scooter injury cases across the United States and found that head injuries accounted for around 28 percent of all injuries, with fractures and contusions making up the bulk of the remainder. Studies from Lund University in Sweden and the German Social Accident Insurance (DGUV) have produced injury mechanism analyses for e-bike crashes, identifying specific collision scenarios that standard certification tests do not simulate. The gap between what lab certification tests and what real-world operation produces is where the argument for ongoing, in-service monitoring technology becomes most compelling.

Regional and International Regulatory Frameworks: A Landscape That Is Still Being Drawn

Outside Europe and the United States, micromobility regulation varies enormously by jurisdiction. In Singapore, e-scooters used on public paths must be registered with the Land Transport Authority, comply with a maximum weight of 20 kg, a maximum speed of 25 km/h, and carry an LTA-approved label. The Fire Safety and Shelter Department has separately introduced mandatory fire safety standards for PMD (personal mobility device) batteries following a series of residential fires. Australia regulates micromobility at state level rather than federally, creating a patchwork where Queensland, Victoria, and New South Wales each have different speed limits, footpath access rules, and registration requirements for e-scooters. France took the step in 2023 of banning rental e-scooters from Paris following a city referendum, a politically significant outcome that illustrated the degree to which public tolerance for shared micromobility is conditional on visible safety standards and enforcement. Paris had been one of Europe's largest shared scooter markets, with over 15,000 rental devices on its streets at peak. The ban has been studied closely by other European city authorities weighing permit renewal decisions. The European Commission has been working toward greater harmonisation through its Sustainable and Smart Mobility Strategy, which targets 100 European cities deploying smart urban mobility by 2030. The Commission's Urban Mobility Framework, adopted in 2021, explicitly addresses micromobility integration and calls on member states to develop clear regulatory frameworks covering vehicle standards, infrastructure, and behavioural rules. However, the Commission's competence in this area is limited by the principle of subsidiarity: road traffic rules remain primarily a member state competency, which is why the speed limit for e-scooters varies from 20 km/h in Germany to 25 km/h in France (before the ban), 20 km/h in the Netherlands for most path types, and higher limits in some Scandinavian contexts depending on road classification. For fleet operators and OEMs, this regulatory fragmentation is not merely an inconvenience. It creates genuine product development complexity: a vehicle platform that meets German StVZO requirements, French national standards, UK trial permit conditions, and Singapore LTA certification simultaneously requires engineering decisions to be made early in the design process, before any single market's requirements are finalised. This is one of the structural reasons why compliance cannot be treated as a post-engineering checkbox.

Rider behaviour sits at an intersection between regulatory compliance and safety outcomes that is often underweighted in technical discussions. The evidence is consistent: a significant proportion of serious micromobility injuries involve behaviours that no vehicle certification standard can address directly, including riding on pavements where prohibited, riding at night without lights, riding under the influence of alcohol, and failing to wear helmets where they are recommended or required. A study from Transport for London published in 2022 found that in recorded e-scooter collisions in London's trial zones, 37 percent of riders involved were not wearing helmets, and a meaningful share of incidents occurred outside designated riding zones. Helmet mandates vary by jurisdiction: Germany requires helmets for speed pedelecs above 25 km/h but not for standard e-bikes or e-scooters; Australia's rules vary by state; the US has no federal helmet requirement for e-scooters though many cities have local ordinances. The policy tension is well-documented: mandatory helmet laws have been shown in some contexts to reduce injury severity but also to reduce uptake of shared schemes, reducing the modal shift benefits that make micromobility attractive to city planners in the first place. This does not mean helmet safety is irrelevant to compliance technology, but it does mean that the relationship between regulation and safety outcomes is not simply additive.

Technology Solutions for Monitoring and Enforcing Micromobility Safety Compliance

The gap between static certification and dynamic, real-world safety performance is where technology solutions have the most distinctive role to play. Fleet telematics has been the dominant approach to date: GPS-based geofencing to enforce speed limits and no-ride zones, accelerometer data to detect falls or collisions, and cellular connectivity to relay vehicle status to fleet management dashboards. Providers including Superpedestrian, Spin, and Tier Mobility have built substantial telematics platforms that allow fleet operators to monitor vehicle health, rider behaviour patterns, and compliance with operational permits in near real time. These systems have genuine value, but they operate reactively and at the network level rather than on the vehicle in the moment of a hazard event. The emerging technology frontier is on-device inference: running machine learning models directly on embedded hardware mounted on the vehicle itself, without routing sensor data to a remote server for processing. This approach, broadly categorised as edge AI or TinyML depending on the computational constraints involved, changes what safety systems can do. A computer vision model running on a vehicle-mounted processor can detect an obstacle, a pedestrian stepping off a kerb, or a vehicle door opening in front of the rider in tens of milliseconds. The latency involved in sending a camera frame to a cloud server, running inference, and returning an alert is measured in hundreds of milliseconds to seconds, which at 20 km/h represents the difference between a useful warning and an after-the-fact log entry. Near Human's edge AI system for micromobility is built around this core constraint: the physics of urban riding at speed mean that safety-relevant decisions cannot wait for a network round-trip. Their system runs computer vision inference on embedded hardware directly on the e-scooter or e-bike, detecting hazards and protecting riders in real time. This is not a generic computer vision SDK adapted from another application domain; it is purpose-built for the specific sensor configurations, computational budgets, and hazard categories relevant to micromobility in shared urban environments. For fleet operators, this matters for compliance in a specific way. Regulators and city permit authorities are increasingly asking not just whether vehicles meet static technical standards, but whether operators can demonstrate active safety management over the vehicle's operational life. An edge AI system that logs hazard detection events, generates safety incident data, and provides evidence of active monitoring gives operators an auditable trail that static certification cannot provide. This is analogous to the trajectory that automotive ADAS (advanced driver assistance systems) followed: what began as a differentiating feature became an expected component of any serious safety compliance claim, and is now increasingly mandatory in new vehicle categories under EU General Safety Regulation 2019/2144.

Sensor fusion is a related technical area that is becoming increasingly relevant to compliance monitoring. Rather than relying on a single sensor modality, fusion approaches combine data from cameras, LiDAR, ultrasonic sensors, and inertial measurement units (IMUs) to build a more complete picture of the riding environment. For micromobility, the practical constraints are more severe than for automotive: sensor systems must be small, cheap, low-power, and physically robust enough to survive the vibration and weather exposure of daily rental use. This is an area where purpose-built design matters considerably. Automotive-grade sensor fusion systems assume sensor positions, fields of view, and computing resources that do not translate to a 15-kilogram scooter. The same engineering discipline that went into fitting neural network inference into the computational envelope of a microcontroller, the core skill of TinyML practitioners, is needed to make sensor fusion viable on micromobility hardware at fleet scale. Pedestrian and cyclist safety adds another dimension to this technology discussion. In shared urban spaces, the safety of a micromobility vehicle is not only a matter of protecting the rider; it is also a matter of the vehicle's interaction with other road users who are considerably more vulnerable than car occupants. A pedestrian struck by an e-scooter travelling at 20 km/h faces injury risks similar to those from a car at lower urban speeds. Computer vision systems that detect and anticipate pedestrian movement in the vehicle's path, particularly in the high-density, unpredictable environments of city centres, contribute to a safety profile that extends beyond the rider to the surrounding public. For city councils evaluating permit renewals or considering new micromobility tenders, the ability of a fleet operator to demonstrate this level of active pedestrian protection is becoming a meaningful differentiator.

How Cities Are Integrating Micromobility Safely Into Urban Transport Systems

The most successful examples of micromobility integration share a common characteristic: the city authority and the fleet operator agreed on technical and behavioural standards before the vehicles reached the street, rather than attempting to impose standards after scale had made compliance politically fraught. Helsinki's city authority worked with operators to define geofenced low-speed zones around the market square and waterfront areas before the launch of its shared scooter scheme, and embedded speed compliance monitoring as a contractual obligation tied to permit renewal. Zurich has taken one of the more structured approaches in Europe, limiting the number of shared micromobility providers to a small number of operators who meet strict technical and operational criteria, including vehicle certification standards, parking discipline technology, and data sharing with the city's mobility management platform. The data sharing element is increasingly central to urban integration strategies. City authorities want aggregated, anonymised trip data to understand how micromobility interacts with their wider transport network, where parking compliance is poor, where speed violations cluster, and how micromobility users transfer to and from public transport. This creates a commercial and regulatory incentive for operators to build data infrastructure that can supply these feeds, which in turn favours operators whose vehicles carry the on-device hardware capable of collecting and logging relevant events. The UK Department for Transport's e-scooter rental trials, which have been extended several times since their 2020 launch, explicitly require trial operators to submit safety data including collision reports, near-miss events where detectable, and vehicle condition monitoring data. The trials have provided one of the more structured datasets on shared e-scooter safety in a European context, though the UK government has been slower than anticipated in bringing forward primary legislation to create a permanent legal framework for e-scooters on public roads. The Urban Mobility Framework from the European Commission calls for integration of micromobility into the broader Mobility as a Service (MaaS) ecosystem, where users plan, book, and pay for multi-modal journeys through a single interface. For micromobility to function reliably within MaaS platforms, the vehicles themselves need to be dependable enough that operators can make commitments about availability and performance. A vehicle that is pulled from service because a battery fault was not detected early, or that generates a safety incident because its hardware has degraded past operational thresholds, creates ripple effects through the MaaS platform's reliability metrics. This is the operational logic that connects vehicle-level safety monitoring technology to fleet-level compliance and, ultimately, to the city-level transport integration that regulators want to see.

Failure analysis is a component of this picture that the industry is still developing maturity around. In the automotive sector, homologation processes require manufacturers to conduct failure mode and effects analysis (FMEA) and demonstrate that critical safety functions are fail-safe or fail-operational. Equivalent requirements for micromobility vehicles are far less standardised. Some manufacturers conduct internal FMEA processes; others do not. When a braking system fails, a battery management system malfunctions, or a controller glitch causes unexpected motor behaviour, the investigation pathway is not always clear. Regulatory bodies in most jurisdictions do not have the same product recall infrastructure for micromobility that NHTSA operates for motor vehicles in the United States, or that the Vehicle Certification Agency provides in the UK automotive context. This is beginning to change: the EU's GPSR, which supersedes the older General Product Safety Directive from December 2024, introduces stronger market surveillance obligations on national authorities and clearer recall and notification requirements for economic operators. For fleet operators specifically, the GPSR reinforces the argument for comprehensive event logging: if a safety incident occurs and a regulatory investigation follows, operators who can provide detailed data on vehicle condition, usage history, and any detected anomalies are in a significantly stronger position than those who cannot.

Compliance is the floor, not the ceiling: the real work of micromobility safety happens in the milliseconds between a hazard appearing and a rider's ability to respond, and no static certification standard has ever operated that fast.

The arc of micromobility safety compliance technology points toward a future that looks more like aviation maintenance than consumer electronics retail: continuous airworthiness rather than point-in-time certification, with data from every journey feeding back into vehicle health models, predictive maintenance schedules, and regulatory reporting. The e-scooter or e-bike of 2030, in a well-regulated urban market, will likely carry embedded hardware that logs its own condition, detects its own anomalies, monitors its environment in real time, and communicates its safety status to the fleet operator and, via standardised interfaces, to the city authority responsible for its permit. The path to that future runs through the unglamorous work that is happening now: engineers in Bristol and Berlin and Bengaluru writing TinyML inference code that fits inside a microcontroller's memory budget, standards committees in Brussels debating EN 17128 annexes, battery chemists designing cells that degrade more predictably, and city planners writing tender specifications that reward genuine safety investment over the cheapest compliant option. A child crossing a cycle lane in front of a rental scooter at dusk does not care about the regulatory framework the vehicle was certified under. What she needs is the scooter to stop.