In recent years, the rapid evolution of autonomous driving technology has propelled Robotaxi into the limelight as one of the most promising applications for L4 autonomous driving technology commercialization. Tech giants have embraced Robotaxi services, with Baidu and Pony.ai leading the charge, having successfully piloted and commenced commercial operations in numerous Chinese cities. Concurrently, the technological advancements in City NOA (Navigate on Autopilot) are enhancing the driving experience, gradually bringing the level of driving assistance closer to L3.

Introduction

The surge in autonomous driving technology is reshaping the global transportation landscape. As an L4 autonomous vehicle, Robotaxi stands out as a pivotal application in the shared mobility sector, offering benefits such as reduced travel costs, improved efficiency, and minimized human errors. Meanwhile, City NOA technology, a stepping stone towards full autonomy, is progressively being implemented in various cities, becoming an integral part of current intelligent driving applications.

Technical Background and Market Development of Robotaxi

2.1 Core Architecture of L4-Level Technology

L4 autonomous driving technology enables vehicles to operate autonomously under specific conditions without driver intervention. This advanced architecture hinges on the integration of high-precision maps, sensor fusion, computing platforms, and intelligent algorithms.

High-Precision Maps: These maps provide detailed road information beyond traditional navigation maps, including lanes, traffic signs, and buildings. Robotaxi leverages high-precision maps for precise location determination and route planning, essential for navigating complex urban environments.

Sensor Fusion Technology: Robotaxi employs a multitude of sensors, including lidar, cameras, and millimeter-wave radars, to perceive the surrounding environment through multi-source data fusion. Lidar constructs 3D environmental models, cameras identify visual information like traffic lights and license plates, while millimeter-wave radars offer reliable distance measurements even in adverse weather conditions.

Computing Platform: The "brain" of Robotaxi is a high-performance computing platform that integrates sensor data to perform real-time environmental analysis, decision-making, and route planning. Modern computing platforms utilize AI chips for parallel computing, capable of processing intricate perception data and executing intelligent driving algorithms instantaneously.

Intelligent Algorithms: Leveraging deep learning and machine learning, Robotaxi's autonomous driving algorithms continually enhance their perception and decision-making abilities by analyzing vast amounts of traffic scenario data. These algorithms not only determine driving routes in real-time but also adeptly handle emergencies such as pedestrians crossing and sudden stops.

2.2 Market Status and Policy Drivers

China has adopted proactive policies to foster the commercialization of autonomous driving, particularly in Robotaxi applications. By issuing test licenses and planning intelligent infrastructure, the government has significantly accelerated industry development.

Beijing has issued hundreds of autonomous driving test licenses and completed the construction of intelligent road facilities covering over 600 square kilometers. Cities like Wuhan and Shanghai have gradually opened test sections for autonomous driving and supported Robotaxi commercial operations. Companies such as Baidu, Pony.ai, and AutoX have spearheaded these initiatives by piloting Robotaxi services in these cities.

Review of Robotaxi Player Operations

According to Huaxi Securities research, Baidu's Robotaxi service has achieved commercial operation in 11 Chinese cities, with over 6 million orders as of April 2024. Didi, Ruqi Chuxing, Pony.ai, and others are also accelerating Robotaxi implementation in other cities. With further policy liberalization and infrastructure improvements, Robotaxi is poised for rapid market expansion in the coming years.

2.3 Technical Challenges and Solutions for Robotaxi

Despite entering the pilot and commercial operation stage, Robotaxi faces numerous challenges in large-scale deployment, particularly at the technical level.

Complexity of Perception and Decision-Making: Urban traffic conditions are complex and variable, requiring Robotaxi to handle unpredictable situations like pedestrians crossing and vehicles driving against traffic. This necessitates an autonomous driving system with exceptional environmental perception and rapid response capabilities.

Solution: By fusing data from multiple sensors like lidar, cameras, and millimeter-wave radars, comprehensive environmental perception is achieved. Additionally, AI large model technology enables Robotaxi's intelligent decision-making system to adeptly handle complex scenarios through extensive data learning.

High-Precision Map Update Issues: Autonomous driving relies on frequently updated high-precision maps, especially in the face of urban road changes such as new roads and closed sections.

Solution: Real-time map update technology, based on the vehicle fleet, allows vehicles to upload environmental changes to the cloud during driving, enabling real-time map updates. Companies like Tesla are also exploring reducing reliance on high-precision maps through advanced visual perception technology.

Cost Control: High costs associated with sensors and computing platforms pose a significant barrier to Robotaxi commercialization. Currently, key components like lidar remain expensive, affecting the overall vehicle economy.

Solution: Collaborations with supply chain enterprises, like Baidu's partnership with Hesai Technology to launch the affordable lidar AT128, have significantly reduced vehicle hardware costs. As autonomous driving technology matures, companies can further cut costs through mass production.

Evolution and Application of City NOA Technology

3.1 Differences Between Progressive and One-Step Technology Paths

City NOA technology represents L2+ level intelligent driving, enabling vehicles to navigate and drive autonomously in urban environments. Depending on the technical strategies of different enterprises, City NOA development can be broadly categorized into progressive and one-step paths.

Progressive Path: Companies such as Tesla, Xpeng, and Li Auto upgrade existing driving assistance functions through gradual iteration, transitioning from L2 to L3 and L4. Tesla's FSD system continuously enhances autonomous driving capabilities through OTA updates.

The "progressive" path for intelligent driving technology involves multiple iterations.

One-Step Path: Enterprises like Waymo and Baidu Apollo adopt a strategy of directly developing L4+ level technology, aiming for full autonomy in fixed areas or specific conditions. While technically demanding, this path facilitates rapid deployment in scenarios like Robotaxi and unmanned logistics once realized.

The evolution of intelligent driving technology follows two paths: "progressive" and "one-step".

3.2 Technological Breakthroughs and Application Scenarios of City NOA

City NOA's primary application is in complex urban road environments, enabling autonomous driving within a set range. Significant progress has been made in perception, decision-making, and control technologies.

The implementation of domestic City NOA intelligent driving solutions is accelerating.

Perception Level: Traditional NOA technologies relied heavily on high-precision maps. However, with the advent of large model technology, companies are moving away from map dependence. For instance, Xpeng Motors has progressively introduced "mapless" autonomous driving, reducing high-precision map reliance and accelerating city expansion.

Decision-Making and Control Level: The application of AI large models is a key breakthrough in City NOA technology. Tesla's FSD, for example, utilizes AI large models in perception and decision-making. Through the BEV (Bird's Eye View) + Transformer deep learning model, Tesla processes vast amounts of data in real-time for intelligent decision-making.

3.3 User Experience and Market Feedback

With technological advancements, City NOA's user experience has significantly improved. Xpeng Motors' City NGP system covers 243 cities, including Guangzhou, Shenzhen, and Shanghai, offering an L3-level autonomous driving experience. Huawei has also launched a nationwide City NOA-supported intelligent driving solution, with its HarmonyOS system enabling OTA remote updates for the latest driving experience.

Consumer acceptance is growing. Market surveys indicate that NOA-equipped smart cars are becoming an important consideration for car buyers, especially young urbanites where intelligent driving capabilities influence purchasing decisions.

Exploration of Robotaxi Business Models and Profitability Challenges

4.1 Diversified Business Models

Robotaxi business models include independent operation, collaboration with travel platforms, and joint ventures with OEMs. Here are several common models:

Independent Operation Model: Enterprises operate their own Robotaxi fleet, handling scheduling, operation, and maintenance. Baidu's Robotaxi service exemplifies this model, offering control over the entire service chain but with higher operating costs, particularly for vehicle depreciation and maintenance.

Collaboration with Travel Platforms: Some Robotaxi companies partner with existing travel platforms, like Pony.ai's collaboration with Didi. This model leverages the platform's user base and operating system, reducing early market expansion challenges and lowering operating costs.

Collaboration with OEMs: For instance, WeRide.ai establishes joint ventures with OEMs, integrating production resources and autonomous driving technology to enhance vehicle capabilities while reducing hardware costs.

4.2 Exploration and Challenges of Profit Models

While Robotaxi commercial pilots have made progress, profitability remains a challenge. Costs include vehicle depreciation, safety officer salaries, charging expenses, and maintenance. Baidu's Robotaxi service, for example, incurs a daily operating cost of 370 yuan per vehicle, with daily order income insufficient to cover these costs.

Achieving profitability is a key issue for Robotaxi. Reducing hardware costs, increasing vehicle order volume, and lowering safety officer costs are crucial for commercial success.

Reduction in Hardware Costs: As lidar and computing platform costs decrease, Robotaxi manufacturing costs will significantly drop. Hesai Technology's AT128 lidar, priced lower than traditional lidars, will rationalize Robotaxi cost structures.

Improving Operational Efficiency: Optimizing vehicle routes through intelligent dispatch systems, reducing empty running time, and enhancing vehicle utilization rates can boost Robotaxi operational efficiency. Removing or minimizing safety officers will further cut operating costs and enhance profitability.

Technical Collaboration and Future Prospects

5.1 Development of Vehicle-Road Collaboration Technology

With the increasing prevalence of 5G technology and the steady advancement of vehicle-to-infrastructure (V2I) collaboration, the integration of Robotaxi and urban Navigate on Autopilot (NOA) technology will become ever more seamless in the years ahead. V2I collaboration significantly enhances the responsiveness and safety of autonomous vehicles by enabling real-time data sharing between vehicles and road infrastructure. Especially in complex urban scenarios, this technology can furnish more comprehensive road information, aiding Robotaxi and NOA systems in optimizing route planning and mitigating risks.

5.2 Prospects for Future Technological Advancements

Beyond V2I collaboration, the rapid evolution of quantum computing and cloud computing is poised to substantially boost the computational prowess of autonomous driving technology in the coming years. Quantum computing's unparalleled parallel processing capabilities will drastically cut down the computation time required for autonomous driving algorithms, while cloud computing offers extensive real-time data processing support, empowering Robotaxi and urban NOA systems to react even faster.

Conclusion

The swift development of Robotaxi and urban NOA technology signifies that the autonomous driving industry is stepping into a pivotal phase of large-scale commercialization. Despite confronting multifaceted challenges encompassing technology, cost, and regulation, continuous technological advancements and gradual policy relaxation indicate that Robotaxi and urban NOA will achieve remarkable breakthroughs in the near future. Particularly fueled by cutting-edge technologies such as V2I collaboration, 5G networks, and large AI models, the commercialization prospects for Robotaxi and urban NOA are broader than ever, positioning intelligent driving as a cornerstone of the global transportation system in the years to come.

-- END --