German carmaker Mercedes-Benz and Chinese autonomous driving tech firm Momenta have marked a new phase in their China partnership following the debut of the all-electric CLA.
The companies’ collaboration centres on deploying advanced driving and parking systems developed specifically for dense urban environments in the country.
According to the press statement, the Urban Pilot function relies on end-to-end large models trained using extensive real-world Chinese road data.
This allows vehicles to manage scenarios such as multi-lane junctions, narrow streets, roundabouts, U-turns and unexpected pedestrian or object encounters, described as “ghost probes”.
They also outlined a jointly developed parking suite planned for 2025, which they said has achieved a 98% success rate and performs on average 11 seconds faster than earlier iterations.
A new “Tracked Reverse” capability enables drivers to exit dead ends with a single command, with the vehicle autonomously retracing its route while avoiding obstacles along paths up to 0.8m wider than the car.
Looking ahead to 2026, Mercedes-Benz and Momenta said they intend to upgrade both urban and highway pilot assistance using Momenta’s R6 large reinforcement learning model.
The technology is due to appear in additional vehicles, including a new fully electric GLC SUV, while mass production and delivery of a “Memory Parking” feature is also planned as part of a broader rollout.
The companies said the developments extend their existing cooperation in China and are aimed at widening the availability of intelligent-driving functions across Mercedes-Benz’s electric range.
They added that the partnership will continue to focus on advancing safety-led automated driving and parking systems for Chinese customers.
Separately, Mercedes-Benz last week disclosed work on a robotaxi programme built around its new S-Class, involving technology providers and mobility operators in multiple regions.
The manufacturer said the model will serve as the core platform, citing redundant steering, braking, computing and power supply systems.
