Summary

• Wayve, Stellantis and NVIDIA will collaborate on autonomous driving technology, using Wayve AI Driver, Stellantis vehicle platforms and NVIDIA computing systems.
• France plans new autonomous vehicle approval guidance in the first half of 2027, with safety, data protection and technological sovereignty central to its approach.
• A separate Greater Paris pilot involving Bolt and Stellantis is planned from 2028; neither initiative represents an authorised commercial service today.

British autonomous driving developer Wayve has announced a collaboration with Stellantis and NVIDIA aimed at developing autonomous vehicle technology in France, alongside government plans to establish clearer approval requirements for vehicles capable of operating without continuous human control.

The initiative was presented on 23 October at the Assises du Véhicule Autonome, a government-led event bringing together vehicle manufacturers, technology suppliers, transport operators, local authorities and trade unions.

The collaboration will combine Wayve’s AI Driver software with Stellantis vehicle engineering and manufacturing capabilities. It also involves NVIDIA’s Hyperion computing and sensor architecture and its Halos operating system technology, creating a potential development framework extending from processors and software to completed vehicles.

Although the partners intend to support eventual deployment in France and across Europe, they have not disclosed a committed commercial launch date, an authorised fleet or a completed approval process associated specifically with this collaboration.

France’s transport ministry has separately confirmed that it intends to publish national autonomous vehicle approval guidance during the first half of 2027, providing manufacturers and operators with greater clarity about the technical evidence required before vehicles can enter service.

France defines a route towards authorisation

The French government has pursued automated mobility through a national strategy established in 2018, with more than 200 experiments undertaken across the country since then.

Research trials and limited demonstrations have helped developers test technology, but moving towards commercial operation introduces requirements extending beyond the performance of an individual driving system.

Vehicles need to function within defined operating conditions, interact safely with other road users and remain dependable when equipment or environmental conditions change. Operators must also establish procedures for incidents that cannot be resolved automatically.

The planned approval guidance will be developed under the Centre national de réception des véhicules, France’s national vehicle approval authority. The government intends to use a progressive approach in which safety is evaluated through development and monitored over the vehicle’s operating life.

This could provide clearer expectations for suppliers preparing technical evidence, although publication of guidance will not itself authorise any individual vehicle or commercial service.

France is also participating in European efforts to harmonise automated mobility arrangements. In October, it helped launch the Pact for Europe coalition of transport authorities, building on earlier cooperation concerning the testing and introduction of automated services.

Such coordination could reduce uncertainty for companies working across different national markets, particularly where a vehicle intended for use in one country must undergo additional assessment before operating elsewhere.

Nevertheless, national operating conditions will still differ. Road infrastructure, traffic behaviour, local service requirements and the legal responsibilities of operators all influence the evidence needed for deployment.

Three suppliers address different technical layers

Wayve is developing an AI Driver intended to interpret driving environments and determine how vehicles should respond. Its approach seeks to use AI capabilities across different vehicle platforms rather than construct a separate driving system for each individual model.

Stellantis contributes its L4-Ready vehicle platforms and the engineering, manufacturing and integration capabilities needed to incorporate autonomous technology into road vehicles.

NVIDIA provides computing and sensing technology through Hyperion, alongside software associated with the operation and development of automated driving systems.

Combining those capabilities creates an opportunity to evaluate the relationship between software behaviour and the vehicle hardware on which it depends.

However, successful integration requires more than compatibility between computing equipment and control software. Steering, braking, electrical systems, sensors and communications must operate reliably together, while engineers need to understand how failures in one component affect the remainder of the system.

Level 4 automation refers to driving systems capable of performing the driving task without continuous human intervention within defined operating conditions. Those conditions may restrict where a service operates or the weather, road types and traffic circumstances it can manage.

A vehicle designed to support level 4 automation is therefore not necessarily approved or capable of operating without a driver in every environment.

The partners must establish how the AI Driver behaves under the operating conditions for which it is intended, while demonstrating that the complete vehicle can respond appropriately to unexpected events.

Public transport integration adds complexity

France wants automated mobility to complement existing transport networks, particularly in areas where conventional services are limited or connections between different modes of transport remain inconvenient.

The government has identified potential applications in rural and suburban mobility, connections with railway networks and urban goods distribution.

Such services require cooperation between technology developers and the organisations responsible for operating transport. A vehicle may be technically capable of following a route but still require arrangements for passenger assistance, maintenance, incident response and communication with local authorities.

At the October event, several separate mobility projects were also presented. Île-de-France Mobilités outlined plans for a public transport pilot beginning in early 2028, while other proposals involved local taxi services and automated vehicles.

The Métropole du Grand Paris, Bolt and Stellantis announced a distinct project to test autonomous mobility services from 2028, initially in up to five municipalities, with potential expansion to approximately 30 by 2030.

That project identifies Pony.ai as a potential supplier of autonomous driving technology and also refers to NVIDIA’s computing and safety systems.

Although Stellantis and NVIDIA are connected to both initiatives, the Greater Paris pilot must not be presented as a confirmed deployment of Wayve’s AI Driver. The partnerships involve different participants and commercial arrangements.

The Greater Paris project also remains subject to approval and independent assessment, with its proposed expansion dependent on the results of initial testing.

Safety evidence extends beyond simulation

Autonomous driving developers use simulation, controlled testing and public-road trials to evaluate how vehicles behave. Each method provides different evidence, while none can reproduce every situation likely to occur during commercial operation.

Simulation can expose systems to a wide variety of scenarios without endangering road users, but its value depends on how accurately those scenarios represent real conditions.

Physical trials provide information about actual vehicle behaviour, although the number of situations encountered during testing is limited by time, geography and operating circumstances.

Regulators consequently need evidence that the complete system performs reliably within its stated operating boundaries, including procedures for circumstances in which those boundaries are exceeded.

France’s intended approach includes monitoring safety throughout the vehicle’s operating life. This recognises that automated driving software and supporting systems may change after a vehicle has entered service, requiring continuing assurance rather than a single initial assessment.

Data handling introduces additional responsibilities because autonomous vehicles can collect extensive information about their surroundings, journeys and passengers.

The French government has placed technological sovereignty and European control over relevant data among its policy priorities. Those objectives may influence where information is processed, the contractual arrangements governing access and the degree of control operators retain over software functions.

However, European operation does not automatically establish that all components or data processing arrangements are sovereign. Vehicles may depend on international technology suppliers, cloud infrastructure and software components whose access and maintenance arrangements require separate examination.

Commercial services require operating partners

Wayve, Stellantis and NVIDIA have described their collaboration as part of a wider effort to establish France as a European centre for autonomous mobility development.

The proposal builds on existing technical relationships between the companies, but the latest initiative also brings national policy, transport operations and local stakeholder engagement into the development process.

The parties have indicated that engagement with trade unions, authorities and mobility stakeholders will be led by local partners responsible for operating transport services.

Those arrangements recognise that introducing autonomous vehicles changes more than the technology used to drive. It also affects employment, passenger assistance, operational responsibility and the integration of new services with existing transport networks.

European vehicle manufacturers and mobility operators are already pursuing several approaches to autonomous driving, as Techopia reported in its earlier examination of NVIDIA’s European automotive programmes.

The French collaboration adds a national development and regulatory dimension to those technical efforts, although the steps needed for authorised operation remain substantial.

The next significant policy milestone is the publication of France’s approval guidance during the first half of 2027. For the companies involved, commercial progress will depend on validated vehicle performance, completed integration work and transport operators prepared to introduce services under the resulting safety and regulatory requirements.