Mapbox has revealed new AI-powered features for its Navigation SDK, which equips OEMs to provide alternatives to smartphone mirroring systems, offering AI-enabled navigation features, integration with in-vehicle systems, and interfaces to third-party services.
The Mapbox Navigation SDK ships with MapGPT, a unified AI-powered voice assistant that facilitates conversations about location that are more natural and actionable. The assistant integrates deeply into the vehicle, allowing drivers to control their vehicle’s navigation, entertainment services, autopilot, and climate control systems, while to accessing capabilities from third-party apps including OpenTable, The Weather Company, and TripAdvisor.
The SDK also includes Mapbox for EV, an EV route planning system featured in new EVs from BMW and Mini. It similarly integrates with vehicle battery systems to monitor personalized energy consumption patterns and intelligently forecast range. Partnerships with charge point operators allow Mapbox for EV to suggest charging stations based on real-time availability and can facilitate secure payment processing. Drivers benefit further from being directed to functional and available chargers, with the ease of a single interface for managing payments across different charging platforms. Mapbox for EV proactively learns from drivers, with data recorded on charger availability, compatibility, and performance every time an EV is plugged in.
Within the field of navigation, the SDK readies Mapbox 3D Live for integration with vehicle sensors, the Autopilot Map, and MapGPT to deliver an SAE L2+ driving experience. At the same time, the upgraded SDK also SDK integrates new models for live and expected traffic to provide more precise estimated drive times, and better routes for drivers in congested areas.
Mapbox Traffic builds on a large network of Mapbox-enabled applications to provide up-to-date insights on congestion, closures, and changes to the road network. The new AI-enhanced traffic models adjust for regionally-specific driving patterns by learning from millions of comparisons of estimated drive times against actual drive times.