News Center



Home News Center News

Technological Innovation Trends & Business Model Challenges of Intelligent Driving

Oct 28-2024   



This seminar brought together a number of faculty members from the National School of Development at Peking University and Peking University alumni to discuss the evolution of intelligent driving technology and the business challenges it presents.

A Historical Overview of the Development of Intelligent Driving Technology

Su Tan, Vice President of Microsoft China, Head of Microsoft's Automotive Business Unit, Peking University Alumnus

 

The first generation of commercially available intelligent driving technologies was mainly led by foreign companies such as Bosch, Delphi and Continental. These companies achieved notable commercial success by integrating forward-facing cameras and basic driver assistance functions. These technologies continue to account for a significant portion of the market due to their cost effectiveness and extended product lifecycle.

 

The development of second generation intelligent driving technologies has its origins in deep learning driven high precision mapping and autonomous driving systems. However, significant cost increases have made commercialisation more challenging across the industry.

 

In 2023, Tesla changed the technology paradigm for intelligent driving through end-to-end large scale models such as the Bird's Eye View model, which significantly improved the overall performance of the system. In addition, Tesla has achieved a new fusion of perception and planning by replacing traditional lidar technology with deep learning models.

 

Overall, while new technologies continue to emerge, the lifecycle of each generation is becoming shorter, R&D and data costs are increasing significantly, and commercial return models are becoming more uncertain. It is clear that identifying a viable business model in this high-cost, high-risk industry has become a critical challenge.

 

Roundtable Forum

1. The dual challenges of unpredictability and rapid iteration

The current unpredictable and rapidly changing technological variability in the autonomous driving and large scale model industries presents companies with a dual pressure that makes it difficult for them to form stable expectations. It is likely that even with conventional technology paths, companies will face challenges in managing the risks associated with architectural change.

 

2. The limitations of end-game thinking in a fast-moving, uncertain technology landscape.

Entrepreneurs who rely too heavily on end-game thinking may find it difficult to adapt to market changes. It is better to avoid prematurely defining the future endgame and instead address challenges step by step.

 

3. Controversy over autonomous driving technology paths

The autonomous driving industry is currently exploring two different technology paths. Tesla's end-to-end solution emphasises the extremes of algorithms and computing power, while the vehicle-road-cloud synergy represents a more systematic approach to building infrastructure. It is important to consider not only the technology itself, but also the broader industry and policy context.

 

4. The long-term value of data

In a rapidly changing environment, data has the potential to provide longer-term value. Compared to the technology itself, the accumulation of data can provide an ongoing source of support for future innovation.

In conclusion, the choice of entrepreneurial direction should not rely too much on end-game thinking, but should focus on market demand and technological development. Data is a valuable resource for entrepreneurship, and industry data plays an important role in future technology iterations. Given the uncertain nature of entrepreneurship, success comes from continuous adaptation and learning, gaining experience through real projects, communicating and collaborating with people, and constantly adjusting the direction.

Technological Innovation Trends & Business Model Challenges of Intelligent Driving

Oct 28-2024   



This seminar brought together a number of faculty members from the National School of Development at Peking University and Peking University alumni to discuss the evolution of intelligent driving technology and the business challenges it presents.

A Historical Overview of the Development of Intelligent Driving Technology

Su Tan, Vice President of Microsoft China, Head of Microsoft's Automotive Business Unit, Peking University Alumnus

 

The first generation of commercially available intelligent driving technologies was mainly led by foreign companies such as Bosch, Delphi and Continental. These companies achieved notable commercial success by integrating forward-facing cameras and basic driver assistance functions. These technologies continue to account for a significant portion of the market due to their cost effectiveness and extended product lifecycle.

 

The development of second generation intelligent driving technologies has its origins in deep learning driven high precision mapping and autonomous driving systems. However, significant cost increases have made commercialisation more challenging across the industry.

 

In 2023, Tesla changed the technology paradigm for intelligent driving through end-to-end large scale models such as the Bird's Eye View model, which significantly improved the overall performance of the system. In addition, Tesla has achieved a new fusion of perception and planning by replacing traditional lidar technology with deep learning models.

 

Overall, while new technologies continue to emerge, the lifecycle of each generation is becoming shorter, R&D and data costs are increasing significantly, and commercial return models are becoming more uncertain. It is clear that identifying a viable business model in this high-cost, high-risk industry has become a critical challenge.

 

Roundtable Forum

1. The dual challenges of unpredictability and rapid iteration

The current unpredictable and rapidly changing technological variability in the autonomous driving and large scale model industries presents companies with a dual pressure that makes it difficult for them to form stable expectations. It is likely that even with conventional technology paths, companies will face challenges in managing the risks associated with architectural change.

 

2. The limitations of end-game thinking in a fast-moving, uncertain technology landscape.

Entrepreneurs who rely too heavily on end-game thinking may find it difficult to adapt to market changes. It is better to avoid prematurely defining the future endgame and instead address challenges step by step.

 

3. Controversy over autonomous driving technology paths

The autonomous driving industry is currently exploring two different technology paths. Tesla's end-to-end solution emphasises the extremes of algorithms and computing power, while the vehicle-road-cloud synergy represents a more systematic approach to building infrastructure. It is important to consider not only the technology itself, but also the broader industry and policy context.

 

4. The long-term value of data

In a rapidly changing environment, data has the potential to provide longer-term value. Compared to the technology itself, the accumulation of data can provide an ongoing source of support for future innovation.

In conclusion, the choice of entrepreneurial direction should not rely too much on end-game thinking, but should focus on market demand and technological development. Data is a valuable resource for entrepreneurship, and industry data plays an important role in future technology iterations. Given the uncertain nature of entrepreneurship, success comes from continuous adaptation and learning, gaining experience through real projects, communicating and collaborating with people, and constantly adjusting the direction.