Motovis's next-generation integrated driving-parking domain controller MagicPilot 2.0 features the Axera M57 automotive-grade SoC, leveraging a lightweight high-performance BEV network and extreme compute resource optimization to achieve seamless 24/7 driving-parking switching. Targeting the mainstream 100,000-200,000 yuan passenger vehicle market, it aims to shift L2+ intelligent driving from an optional add-on to a standard feature.

Chip Selection Strategy: Axera M57's Differentiated Positioning

MagicPilot 2.0's core computing power comes from the Axera M57 automotive-grade SoC. Unlike numerous solutions on the market that adopt Horizon Robotics chips, Motovis chose Axera — a relatively niche but high-performing domestic chip platform — reflecting its strategic thinking on supply chain diversification.

The M57 chip provides MagicPilot 2.0 with the following technical foundation:

• Lightweight BEV network: A high-performance perception network optimized for integrated driving-parking scenarios
• Extreme compute and bandwidth optimization: Maximizing algorithm efficiency within limited hardware resources
• Highly platformized system design: Supporting rapid adaptation for domestic and international models on a single platform

The core value of this chip selection strategy lies in cost control. By choosing a more cost-effective chip platform, Motovis can keep L2+ intelligent driving system costs within the acceptable range for 100,000-200,000 yuan passenger vehicles, thereby driving true democratization of intelligent driving.

Always-On Driving-Parking Integration: Redefining User Experience Boundaries

MagicPilot 2.0's biggest product differentiation lies in its "Always-on" 24/7 integrated driving-parking experience. By optimizing the underlying state machine and overall software architecture, the system enables seamless switching between driving and parking modes, remaining active around the clock without manual intervention.

This design delivers three significant user experience upgrades:

1. Zero learning curve: The system is always online; users don't need to learn when to activate or deactivate it
2. Instant response: The system can enter intelligent driving state immediately in any scenario
3. Full-scenario coverage: A complete chain from departure to parking without functional breakpoints

In contrast, most intelligent driving systems on the market require users to manually activate them in specific scenarios, and driving and parking functions are often handled by different hardware or software modules, creating obvious experience fragmentation. MagicPilot 2.0's "always-on" design essentially transforms intelligent driving from a "feature that needs to be actively enabled" into "infrastructure that is always available by default."

Single-Platform Global Reuse: The Key to Shortening R&D Cycles

Building on extensive global mass-production experience, Motovis has created a highly universal platform architecture for MagicPilot 2.0. The core advantages of this architecture include:

Dimension	Traditional Solution	MagicPilot 2.0 Platformized Approach
Adaptation Cycle	Independent development per model	Rapid adaptation on single platform
Regulatory Compliance	Repeated certification processes	Reuse of already-passed regional certifications
Cost Control	High integration costs	Significantly reduced R&D and integration costs
Model Coverage	Model-by-model rollout	Rapid adoption across entire lineups

This platformization strategy is particularly significant for OEMs. Against the backdrop of accelerating intelligent driving feature iteration, OEMs need solutions that can be rapidly deployed and low-cost integrated, rather than bespoke systems requiring lengthy custom development. MagicPilot 2.0's "building block" platform design precisely meets this need.

Data Closed-Loop Iteration: AI Large Models Drive Continuous Evolution

Another technical highlight of MagicPilot 2.0 is its full-link data R&D closed loop. Driven by AI large models, the system covers the complete process from data governance, intelligent labeling, model training, simulation verification to mass-production iteration:

1. Data governance: Establishing standardized data collection and cleaning processes
2. Intelligent labeling: Using AI assistance to improve labeling efficiency and quality
3. Model training: Continuously optimizing perception and decision models based on large-scale data
4. Simulation verification: Validating algorithm boundary scenario performance in virtual environments
5. Mass-production iteration: Rapidly pushing verified models to production vehicles

The value of this closed-loop iteration mechanism lies in its ability to efficiently optimize capabilities for long-tail scenarios, continuously upgrading the intelligent driving experience on mass-market hardware and ensuring standard features remain cutting-edge. For consumers, this means the intelligent driving capabilities of their purchased vehicles will continue to evolve through OTA updates rather than remaining static.

Target Market and Core Application Scenarios

MagicPilot 2.0 explicitly targets the mainstream 100,000-200,000 yuan passenger vehicle market, guided by the philosophy of being the "top choice for democratized intelligent driving." Its core application scenarios include:

• Highway cruising: ACC adaptive cruise + LKA lane keeping + LCC lane centering for foot-off driving throughout the journey
• Urban congestion: TJA traffic jam assist + ICA integrated cruise assist for automatic start-and-stop in congested areas
• Full-scenario parking: Covering all space types including parallel, perpendicular, diagonal, nose-in, extremely narrow spots, wall-limited U-turns, and mechanical spaces
• Extreme scenarios: One-click adaptation to high-frequency pain points such as extremely narrow spaces, mechanical spots, restricted U-turns, and severe weather

Through this scenario coverage, MagicPilot 2.0 is driving L2+ intelligent driving from an exclusive configuration for premium vehicles to a standard feature for mass-market models. For more intelligent driving technology news, visit EX1000.COM.