At the WNAT-CES 2025 New Automotive Technology Cooperation Ecosystem Exchange, Song Yang, founder and CEO of IMOTIONTECH, shared key insights from his nine-year journey in autonomous driving. He emphasized that integrating algorithms, software, and hardware is a time-consuming process, but once achieved, it enables rapid algorithm iteration.
In an era where AI and the automotive industry are rapidly converging, speed dominates—fast technological upgrades, swift capital inflows, and quick product launches. Yet, Song highlighted the value of "slowness." True speed, he argued, stems from meticulous groundwork. IMOTIONTECH has focused solely on autonomous driving controllers, patiently refining its technology stack and engineering capabilities to lay a solid foundation for future breakthroughs.
The evolution from CNN to BEV and end-to-end technologies in autonomous driving requires deep adaptation and optimization at the controller level. Song likened this to "carving intricate patterns on a walnut kernel," demanding immense patience and precision. However, once the integration is complete, algorithm iteration accelerates significantly.
This dedication to "slow craftsmanship" has yielded tangible results. Song noted, "Over nine years, we’ve achieved several milestones: our controllers are now used in over 100 countries, featured in Top 100 vehicle models, and surpassed one million units in autonomous driving controller shipments by Q3 this year."
Crucially, this deep technical foundation enables cross-industry innovation. Autonomous driving and embodied robotics share highly similar technology stacks. Song explained, "Comparing autonomous driving and robotics, the controller’s underlying architecture is identical. The software layer is largely the same, and motion control algorithms—especially wheel control—are identical. Overall, the overlap exceeds 90%."
While many companies struggle with the "algorithm-controller-mechanics" bottleneck, IMOTIONTECH leveraged its nine-year expertise to launch the world’s first domestically produced robot controller in just two months.
At the forum themed "Long-Term vs. Short-Term," Song discussed the convergence of automotive and embodied AI. Key challenges include: - The timeline for achieving L4/L5 autonomy: Will it be a sudden leap or gradual data-driven iteration? - Bridging the supply chain gap: Automotive production capacity (~30M units/year) dwarfs robotics (~500K units in 2024). - Industry fusion: Autonomous vehicles are essentially robots, but how can these sectors collaborate now and in the long term?
Song remains optimistic about China’s AI potential, citing its robust industrial base and diverse application scenarios. He believes localized innovation—like DeepSeek’s integrated AI advancements—can propel China ahead in AI-driven industry transformation.
IMOTIONTECH’s nine-year focus on autonomous driving controllers (integrating algorithms, software, and hardware) positioned it to pivot into robotics in 2024. Its robot controller, launched this September, already serves over a dozen clients. Applications include autonomous charging—a critical need as driverless vehicles proliferate.
The company’s vertically integrated approach extends to robotic arms and joints, including acquiring a smart joint manufacturer to streamline algorithm-to-hardware development. Its new 7,200 sqm factory boosts production capacity to nearly 2M units annually, supporting both automotive and robotics industries.
Song concluded by advocating for ecosystem collaboration, inviting苏州-based manufacturers to join the Automotive Electronics and Parts Association to advance autonomous driving and robotics collectively.
Comments