专访 || 清华大学车辆与运载学院教授李升波：我们正在推动一条全新的端到端自动驾驶路线

Core Insights - The recent regulatory changes in the intelligent connected vehicle sector aim to enhance safety and set higher standards for products, responding to market and consumer concerns [1][4] - The concept of "intelligent driving equity" reflects the industry's ambition to make advanced features accessible to lower-end models, but safety must remain the priority [2][4] - The distinction between assisted driving and autonomous driving is crucial, as current mass-produced vehicles only offer assisted driving capabilities, requiring driver supervision [3][4] Regulatory Developments - New policies have been introduced, including a notice on product recalls and safety standards for intelligent connected vehicles, indicating a shift towards stricter oversight [1][4] - The government emphasizes the need for accurate marketing and consumer education regarding intelligent driving features to ensure public safety [4] Industry Trends - The industry has seen a shift from aggressive marketing of intelligent driving technologies to a more cautious approach, reflecting the need for safety and reliability [2][4] - The evolution of intelligent driving technology is marked by a transition from rule-based systems to data-driven, end-to-end solutions, enhancing performance and adaptability [5][6] Technological Innovations - The end-to-end approach in autonomous driving leverages neural networks for all system modules, aiming for a direct mapping from perception to control commands [6][7] - China's exploration of end-to-end technology has led to the development of unique solutions that address local challenges, such as data scarcity and computational limitations [8][9] Future Directions - The integration of "vehicle-road-cloud" systems is proposed as a solution to enhance the capabilities of autonomous driving, allowing for better data collection and real-time decision-making [13][14] - The focus on ensuring safety in extreme scenarios is critical, as the consequences of failures in autonomous driving can be severe [16]