[240809] #Tesla #Li Auto #ZF

1. Musk signals delays to robotaxi unveiling for key design change

• Tesla CEO Elon Musk announced a delay in the unveiling of the company’s robotaxi to incorporate a front design change and other features.
• The event, initially planned for August 8, has been pushed to October.
• The delay was confirmed by an unnamed source, with no additional details provided.
• Analysts and investors were not surprised by the delay due to the challenges in developing autonomous vehicles.
• Musk had previously mentioned the robotaxi platform would replace an inexpensive car project and hinted at both Tesla-owned and individual-owned robotaxis.
• Tesla shares rose by following Musk’s endorsement of Donald Trump for the U.S. presidential race, fanning hopes that a potential Trump administration could make it easier for Tesla to secure regulatory approvals for robotaxis and autonomous driving systems.
• Musk had projected mass production of robotaxis without steering wheels or pedals by 2024 but has missed previous self-driving vehicle targets.

2. Self-Driving Three-Wheelers Deliver Fast Food in Houston

• Faction is testing self-driving three-wheelers for food deliveries in Houston, Texas.
• The EVs, based on Arcimoto motorcycles, use Faction’s DriveLink system and TeleAssist remote monitoring.
• The vehicles are currently fulfilling orders for major outlets, including a Chick-fil-A store.
• DriveLink incorporates sensors, thermal cameras, radar, and drive-by-wire tech for automation.
• TeleAssist provides real-time supervision and decision-making from a control hub.
• The EVs, capable of reaching 75 mph, are in early testing and provide real-world curbside deliveries.
• Feedback and data collected aim to optimize navigation algorithms and enhance AV efficiency.
• A Chick-fil-A on Kirby Drive and Southwest Freeway is part of the pilot, offering curbside delivery via the app.
• EVs will initially have human supervisors, with plans to operate without assistance by August.
• Verizon is providing mobile connectivity for monitoring and tracking.
• Faction CEO Ain McKendrick highlights the potential to revolutionize urban transportation and delivery.

3. ZF Unveils ADAS for Commercial Vehicles with Automatic Lane Change System

• ZF unveiled new safety technologies for commercial vehicles on July 8, 2024, aimed at surpassing current regulations with advanced driver assistance systems (ADAS).
• The technologies include a highway assist system for automatic lane changes and a system using sensors and cloud data to estimate road friction and improve control.
• The Truck Trailer Link provides 360-degree high-resolution monitoring of the truck and trailer combination.
• ZF’s ADAS technology reduces accident risk with systems ranging from individual components to complete solutions, meeting the latest General Safety Regulation (GSR) standards.
• The ADAS technology includes software and algorithms to assess surrounding dangers, detect driver drowsiness and distraction, and determine necessary actions.

4. Li Auto advances in self-driving tech, aiming to close the gap with Tesla’s FSD

• Li Auto unveiled its end-to-end (E2E) self-driving technology architecture on July 5, consisting of an E2E model, a vision language model (VLM), and a world model.
• Li Auto’s approach integrates sensor input to driving trajectory output through a single model, contrasting with segmented solutions from Huawei and Xpeng.
• The new technology aims to handle complex urban driving and addresses testing and verification challenges, and has completed prototype verification and real vehicle deployment.
• The E2E model, inspired by Tesla’s approach, eliminates intermediate rule intervention, potentially offering advantages in information transmission and model iteration.
• The architecture includes fast thinking (System 1) for routine tasks and slow thinking (System 2) for complex scenarios, both trained and verified in the cloud using the world model.
• The VLM, with 2.2 billion parameters, enhances the understanding of complex traffic environments and reduces inference time to 0.3 seconds after optimization.
• Li Auto uses Nvidia’s Orin chip for E2E solutions and may need Nvidia’s next-generation chip, Thor, for larger models.
• Testing and verification combine reconstruction and generative simulation to create a virtual environment for autonomous driving system capabilities.
• The segmented E2E navigate-on-autopilot (NOA) solution will be available to users in July, while the full dual solution (E2E and VLM) is not yet ready for delivery.
• The smart driving industry faces challenges in balancing large-scale user experience and technological advancements, requiring strong engineering and technical judgment capabilities.

5. Autotech startup Nullmax launches new tech to tackle self-driving challenges in more human-like ways

• Nullmax unveiled its next-gen autonomous driving technology, Nullmax Intelligence, in Shanghai, designed for more human-like perception.
• The system uses cameras primarily but also incorporates voice, text, and gestures for comprehensive input.
• Nullmax Intelligence uses an end-to-end model for visual perception, scene description, and driving behavior output.
• Founded in 2016 by former Tesla engineers, Nullmax aims for intelligent driving with a focus on visual perception.
• Nullmax Intelligence performs obstacle detection and 3D reconstruction using vision and generates local maps in real time without high-precision maps.
• The system operates with less than 100T of sparse computing power, reducing costs for full-scenario pilot-assisted driving.
• Nullmax uses real and AI-generated virtual data for model development and iteration.
  Nullmax plans to launch full-scenario intelligent driving solutions by 2025 and expand into freight transportation and robotics.
• Nullmax has collaborated with various car companies, achieving revenue of approximately RMB 34 million ($4.7 million) in H1 2024.