A：The operating system is undergoing a huge transformation, starting to possess the characteristic of "burning fuel", where data becomes the fuel, and the operating system is likened to an engine. This change has given birth to the physical AI world, enabling software, especially AI, to change hardware products like cars and interact with real-world objects or people. Physical AI will bring disruptive impacts in the next ten to twenty years.

A：The VLA operating system is a system similar to the cerebellum responsible for coordinating movement in humans, consisting of three core elements: models, computational power, and data. Xiaopeng Motors has experienced the evolution from software rules to algorithms to end-to-end models in the field of autonomous driving. VLA is currently being upgraded and evolved from a powerful model from vision to language to movement, attempting to eliminate the intermediate language translation step and directly transforming vision into movement rules to achieve a more efficient and safer autonomous driving experience.

A：Xiaopeng invested a large amount of resources in VLA, including 30,000 cards of computing power and training costs of over 20 billion. After multiple failures and discussions, they finally found significant progress in the second quarter of 2024. They developed both standard VLA and innovative VA in parallel. The innovative VA significantly reduced information loss and reasoning time, improved frame rate and driving safety, and excelled in responding to various complex scenarios, achieving a new paradigm for mass production of physical world models two years ahead of schedule.

A：The second generation of VIA focuses on vision, imitating human brain cognition, learning, and observing the world without the need for intermediate language translation. It directly converts visual information into motion trajectories, reducing information loss, improving reasoning efficiency, and reaction speed. In addition, by using a dataset of nearly 100 million video clips and self-developed chips, models, and compiler optimizations, industry-leading computational power and parameter volume have been achieved. This allows vehicles to more accurately predict and respond to complex road conditions, such as narrow roads, parking lots, and other scenarios, greatly enhancing urban driving assistance capabilities and highway driving safety.

A：Because many roads in Europe are small, dual-carriageway roads are rare. In this kind of environment, autonomous driving technology cannot operate effectively, for example, when encountering oncoming traffic and needing to reverse or give way in complex scenarios. Therefore, only by perfecting autonomous driving on small roads and in enclosed areas, and without the need to learn traffic regulations from around the world, can global autonomous driving be achieved. Xiaopeng has decided to fully open source the second generation VI autonomous driving technology, including chips, EEA architecture, models, etc., and welcomes cooperation from Chinese and global vehicle manufacturers or TOI vendors. Several manufacturers are currently in talks.

A：In wet and slippery road conditions and various abnormal scenarios, Xiaopeng's new generation second generation VA takeover rate is only one fifth of FSD, and in complex small road tests, Xiaopeng's system did not encounter any safety issues, and completed the test in less than ten minutes. Recently, FSD released version 14, and we plan to conduct comparative tests in the United States next month.

A：Yes, for example, vehicles can recognize and respond to situations where someone is signaling for the vehicle to stop, or even automatically creep slowly forward when the traffic light changes. These functions have not been specifically developed by us, but they are already appearing in vehicles.

A：Xiaopeng is developing a feature called "Super LCC and human co-driving", which allows users to use automatic driving without relying on navigation, and can assist in automatic driving smoothly on straight roads. It is expected to be rolled out to Chinese and global users in the first quarter of next year.

A：Volkswagen will be the first customer for Xiaopeng's second-generation V2A, and at the same time, the Turing AI chip has also secured a designated position with Volkswagen. Both parties will continue to cooperate to promote the development of autonomous driving in China and even globally.

A：VOM has efficient voice control capability for local internet, can cooperate with automatic driving systems to achieve low latency interaction, support mixed language commands, and is developing multilingual capabilities to enable vehicles to communicate with users in local languages around the world.

A：Xiaopeng believes there are two forms of autonomous driving cars in the future: one is the robot taxi in a full-sharing mode; the other is a privately purchased family-shared car that users can either drive themselves or occasionally drive. Xiaopeng plans to launch three autonomous rental cars in 2026, using self-developed software and hardware to solve the challenges faced by autonomous driving technology, and attempting to apply some of the technology to the consumer market.

A：This L-class experience car for ordinary consumers is called Xiaopeng Novo. It is born from the same source as Novo Taxi, with the same hardware configuration, similar software capabilities, and safety redundancy design. Xiaopeng Novo will be equipped with a full safety redundancy system, have 3000 sets of effective computing power, and have two sets of autonomous driving modes to meet different scenario needs: one focusing on experiential agility, and the other focusing on safety and non-takeover control. In addition, under the circumstances permitted by laws and regulations, the possibility of exploring more innovative models, such as cloud-based safety officer services, may be considered. It is even envisioned that in the future, it could perform daily travel tasks, such as picking up family members for shopping or taking children to and from school.

A：Xiaopeng Motors hopes to achieve win-win results by sharing revolutionary hardware and software products with global partners. They develop software, manufacture hardware, and open up SDKs, allowing partners to purchase and operate Xiaopeng's robot taxi vehicles at the town or city level, creating a unique robo taxi ecosystem. Additionally, Xiaopeng Motors has established a global ecological partnership with Amap, and it is expected that they will promote robot taxi services on a large scale through ecological partnerships in the future, not limited to China, but widely applied globally.

A：Xiaopeng Motors believes that if robots do not imitate humans, it will be difficult to obtain valuable data from humans for training; all scenarios are designed for humans, so humanoid robots are more easily applied in general. In addition, humanoid robots are more easily accepted commercially and can more easily coexist harmoniously with human living environments. The new generation of Allen robots adopts a design similar to humans, with flexible bones, muscles, and skin, capable of performing basic human movements, and can be customized in terms of body shape and gender, and even in the future can enter homes to become companions and colleagues in daily life.

A：The dexterity of the hands is the hardware that is most difficult to mass produce on Alan. With up to 22 degrees of freedom in the hand joints, the number of joints in a single hand is extremely rare in the industry. In order to meet the requirements of the human body's most delicate organ, we have specially designed the smallest sloped joint, with a diameter of only 16 millimeters, which is smaller than 60% of a coin, and achieving precise operation of 22 joints in such a small space is a huge technical challenge.

A：Compared to semi-solid batteries, all-solid-state batteries are safer, especially for applications in households, factories, and company environments where safety is of utmost importance. Humanoid robots, due to their unique environment and high safety requirements, have become the ideal carrier for the mass production and implementation of all-solid-state batteries.

A：The new generation robot Xiaopeng is equipped with three Turing AI chips, with a computing power of 2250 tops, making it the most powerful integrated computing power in the world. At the same time, through the combination of large physical world models and the three major models of VLT, VOA, and VEOM, especially the newly added VOT model, it has realized a high degree of intelligence and comprehensive ability to adapt to various environments and tasks.

A：The VLT acts as the core engine of autonomous robot behavior, equivalent to the "slow brain" of the robot, responsible for thinking, decision-making, and task decomposition, determining how the robot will act next in a real physical environment, including a comprehensive combination of actions, language, and more.

A：Compared to a car with only one engine, a robot has 82 joints, which can create countless combinations of movements. It is a huge challenge to control and coordinate the whole body from feet to hands, waist, shoulders, and neck. We are committed to integrating full-body control, hand-eye-foot actions through the second-generation VI model, establishing the first data factory with embodied intelligence to obtain massive long-term data, ensuring that robots can generalize learning and achieve long-term high-speed growth.

A：We are discussing the need to add a fourth rule of robotics - not to disclose human privacy data, and propose an active security protection rule, similar to the AEB system in cars, to prevent robots from inadvertently harming human and property safety in any situation, especially actively protecting the surrounding safety during movement.

A：Compared to prototype development, the difficulty of mass production robots has increased by 100 times. The main challenges include: a new mode where software drives hardware, strict automotive-grade quality requirements (such as the design of all solid-state batteries), the necessity of self-developed integrated innovation across the stack, and an integrated solution for highly coordinated software and hardware. Xiaopeng has formed a team of over 1,000 people, spanning multiple research and development centers and departments to jointly develop and integrate solutions to address these mass production challenges.

A：Xiaopeng believes, after in-depth research, that direct screwing by robots in Chinese factories is not the most priority scene due to significant cost differences. In household scenarios, it is difficult to achieve in the short term due to the complexity of safety and generalized learning in family environments. Therefore, Xiaopeng has chosen guidance, shopping assistance, and patrol guidance as the main development direction at the current stage, which are basic and easy-to-implement service scenarios, and plans to gradually promote robots entering exhibition halls, sales stores, and other commercial places for work.

A：Xiaopeng Motors aims to achieve mass production of high-level humanoid robots by the end of 2026, which will be a significant challenge and success for Xiaopeng's technological capabilities.

A：All AI car teams of Xiaopeng Motors will collaborate with robot teams to jointly address software and hardware issues, in order to accelerate the high-quality and high-security scale production of integrated advanced capabilities. They hope to become one of the first companies in China, if not the first, to achieve this goal next year.

A：Yes, Xiaopeng's Iron will open its SDK to developers worldwide, through extensive collaboration and secondary development, to enhance its support for more scenarios and capabilities, creating infinite possibilities.

A：Baosteel Group will become the ecological partner of Xiaopeng Robot's IN project. In future applications, Xiaopeng's robots will explore industrial inspection scenarios at Baosteel and help upgrade Baosteel's intelligent manufacturing.

A：The all-new generation of Xiaopeng humanoid robots has the most human-like hardware form, expected to achieve a technological breakthrough in full solid-state batteries by 2026 and begin mass production. Technology Daily reported on the progress of the second generation of VA large-scale models driving autonomous driving technology, with the expectation of launching robots with high-level intelligent configurations and higher-level intelligent vehicle models in 2026; meanwhile, Xiaopeng showcased the entire process from dream to realization in the field of low-altitude travel, emphasizing Xiaopeng's focus on exploration and development in the field of physical AI, aiming to become a leading global provider of intelligent travel solutions.

A：Xiaopenghuitian's R&D team has launched a new generation of flying car A868 in the past 12 years. This model features high efficiency and flexibility, and has now entered the test flight phase. It is expected to drive the development of low-altitude economy towards two directions: land-based aircraft carriers and vertical takeoff and landing. Xiaopenghuitian flying car strictly adheres to the highest flight safety standards, adopting a six-axis, six-propeller and dual-electric duct safety configuration to ensure safe flight and landing in case of any component failure. The first flying car is planned to be launched in the third quarter of 2026, and 6S flagship stores will be established nationwide to provide one-stop services such as display, sales, delivery, training, and services.

A：At today's press conference, Xiaopeng introduced some innovative achievements in the fields of AI and physics, and announced that at the Xiaopeng X9 Kunpeng Super Range Extension Technology release conference on November 6th, a new era of dual-energy sources, from pure electric to pure electric and super range extension, will be showcased. The press conference concluded by mentioning the Xiaopeng X9 Kunpeng Super Range Extension Technology release conference on November 6th, where Xiaopeng's major breakthroughs and innovations in the field of physical power will be revealed.

A：The second generation Xiaopeng robot adopts a design with full-body soft covering, with a more human-like appearance and a sense of safety, suitable for a wider range of human work scenarios. Its body can be customized, the skin material is soft and is expected to incorporate more sensor functions, achieving more flexible application requirements.

A：The second generation of robots can indeed change their skin, using materials that ensure both visual comfort and flexibility of movement. In the future, more sensor capabilities may be integrated into the skin, such as touch and temperature sensing, but the current core concept is to achieve a customizable robot appearance. Regarding the question of when the second generation of robots will enter households everywhere, although no specific time frame was given, it was indicated that this process will be faster than expected by industry professionals.

A：The displayed flying car is the first mass-produced modular flying car, consisting of a road module (extended-range new energy vehicle) and a flying module, solving the difficulties of traditional aircraft in transportation, storage, operation, and learning, allowing users to easily experience outdoor flying. At the same time, the flying car is equipped with a single joystick control system, easy to operate and user-friendly, and has a redundant safety system to ensure flight safety.

A：The land carrier, as a new energy vehicle, adopts extended range power system, with strong endurance ability, supports regular C driver's license driving, and incorporates Xiaopeng Motors' latest autonomous driving technology. The vehicle has six-wheel drive and rear-wheel steering function, with a small turning radius, flexible control, and is equipped with streaming media rearview mirrors, providing a good visual environment.

A：When fully fueled and fully charged, the aircraft can be refueled up to six times, and the charging speed is very fast. It can go from 30% to 80% in half an hour.

A：This flying car is currently mainly used for short-distance flying entertainment experiences, with a range of approximately 15 minutes to 20 kilometers. The aircraft has an automatic separation and integration function with the vehicle, through a one-click separation and slide rail system, the aircraft can be stored in the trunk in less than four minutes, the entire process is precise and does not require manual operation.

A：For friends interested in low-altitude economy or low-altitude flying, they should be the first to learn about the land aircraft carrier of Xiaopeng Huitian, as it is a globally unique modular flying car that can achieve flying freedom.

A：The Xiaopeng X9 super extended-range version has a comprehensive range of 1602 kilometers and a pure electric range of 452 kilometers. It is equipped with the world's longest-range 7-seat design and has seamless switching extended-range technology, achieving long-range endurance while maintaining a pure electric driving experience.

A：Our X9 extended-range model will bring some VRA functions and upgrades in onboard chip capabilities, and there will be an official pre-sale release event tomorrow night at 7 p.m. We will introduce in detail these new features, as well as the exterior and interior color changes of the vehicle. The X9 extended-range model will be officially unveiled at the release event tomorrow, and we will explain in detail the efforts we have made in the extended-range field, explain why we have chosen to switch from pure electric to extended-range, and answer any relevant questions about extended-range models.

A：Xiaopeng AI Day is an annual event aimed at showcasing our company's progress, efforts, and breakthrough research in the field of technology over the past year. This year's AI Day will cover a wide range of research directions, from low-altitude flight to household robots, reflecting Xiaopeng Motors' continuous efforts in exploring the future of travel and lifestyle.

A：Xiaopeng Motors aims to become a global intelligent mobility company. We have conducted extensive research and exploration in the field of AI and through continuous practice, gradually realized the future mobility and lifestyle modes that we showcase.

A：Audiences can leave messages in the barrage or comments section, sharing their views on Xiaopeng Motors' exploration in the field of technology, and making suggestions on what new areas Xiaopeng Motors could pioneer in the future or what surprises they could bring. Emergence refers to the phenomenon of generating new data and capabilities after processing a large amount of data, similar to Xiaopeng Motors' technological innovation in the field of autonomous driving. In the past 11 years, Xiaopeng has continuously explored and self-developed, advancing some innovative achievements from full-stack self-research to cross-domain integration, such as new products and new scenarios like flying cars, embodying the concept of emergence.

A：Today's Xiaopeng Technology Day will release four important AI-based technology applications. This is a very important technological leap in the history of Xiaopeng Motors, which will help drive the company into a new stage of development.

A：Xiaopeng's second generation VIA focuses on vision as its core, training on a large data set of nearly 100 million video clips, paying special attention to corner cases and long-tail data. Through deep learning, it seeks to understand and deduce the physical world, achieving the transformation from physical world models to reasoning to generation. Additionally, Xiaopeng utilizes the enormous computational power provided by Alibaba Cloud, as well as optimizations at the chip, operator, and model levels, resulting in a more than tenfold increase in model parameters and greatly improving the performance of the autonomous driving system.

A：Since 2024, Xiaopeng has invested 30,000 cards of computing power and spent over twenty billion in training costs, forming two teams for research and development. Despite experiencing many failures and hardships, they finally made a major breakthrough one day in the second quarter of 2024, turning from unusable to usable in specific scenarios, thus launching the next generation of VERA.

A：Xiaopeng directly converts the visual world into motion trajectories by building a world model with vision at its core, greatly reducing information loss and improving reasoning efficiency. This physical world model enables vehicles to perceive, learn and observe the world like humans without the need for an intermediate language translation process, thereby achieving a safer, smoother, and more powerful autonomous driving system.

A：The second generation VIA will be launched in the small road park NGP, solving complex driving scenarios in small roads and parks, increasing the average take-over miles by 13 times, and providing a smoother experience from parking space to parking space. This technological breakthrough not only improves urban assisted driving capabilities tenfold, but also increases highway assisted driving safety by 100 times, allowing Xiaopeng to take a big step forward in autonomous driving worldwide.

A：In complex road testing scenarios, Xiaopeng's second-generation VIA has a take over rate of only one-fifth of FSD. Under the same conditions, Xiaopeng's new system can complete the task in just about ten minutes, and there have been no safety supervision issues. This indicates that Xiaopeng's second-generation VIA performs exceptionally well in wet, slippery, and abnormal scenarios.

A：We plan to fully launch this feature to users in China and globally next year. Recently, Xiaopeng's automatic driving upgrade has been put on hold because the team is fully switching to the new generation VIA, and media test drives will be open this month, with pioneer user co-creation experiences opening at the end of next month. At the same time, the European series will push the all-new VA feature in the first quarter of 2026, and for max version users, we are evaluating how to migrate the new model to existing models.

A：Yes, Xiaopeng will fully open source the second generation UAI large model from chip to EEA architecture, model, etc., and welcome global vehicle manufacturers and TOO manufacturers to cooperate. The technology of the second generation V2A will be first applied to the new mass-produced cars of Volkswagen.

A：Xiaopeng believes that there are two types of autonomous driving cars in the future: one is the robot taxi in the fully shared model; the other is a car that individuals can purchase and own with L4 hardware and software experience, which can be shared for use within the family. Xiaopeng plans to launch three models of Robo-Taxi in 2026, and is committed to solving the high cost, legal restrictions, and path planning issues faced by autonomous driving cars in order to promote their larger-scale commercial application. At the same time, Xiaopeng will also explore launching L4-level experience cars for ordinary consumers, providing users with a vehicle experience that is close to Robo-Taxi in terms of hardware configuration and software capabilities.

A：We hope to develop revolutionary hardware and software products with global partners including China, self-develop software, self-manufacture hardware, and open SDK. The future vision is that Chinese small town distributors or partners can purchase and operate 30 Xiaopeng robot taxis, forming a new local transportation model. We hope to jointly build a robo taxi ecosystem in China and even globally through ecological cooperation, which will also be a new entrepreneurial direction for those interested in starting a business.

A：I am very happy to announce that Gaode will become Xiaopeng's Robo Taxi's first global ecological partner. In China, Xiaopeng's Robo Taxi will be integrated into the Gaode aggregate taxi platform, providing users with efficient and seamless travel experience. On a global scale, we will join hands to go abroad and promote China's leading robot taxi service to the world stage. This cooperation is not only an upgrade, but also a key step for both parties to explore AGI together.

A：In the past seven generations of research and development, Xiaopeng initially tried a quadruped design, but it was found through research and development that quadruped robots faced problems such as lack of hands and difficulty turning in household environments. Although robots of various forms could be applied in different scenarios, Xiaopeng believed that the most attractive choice was a humanoid design that is easier to obtain real data, generalize, and commercialize. The new generation of robots has chosen a form closest to humans, and plans to launch three mass-produced robot taxi models with full-stack self-research in 2026, as well as a high-end self-driving version.

A：The all-new generation of robots have a humanoid height, flexible skeleton, extremely tight muscles, and soft skin, capable of performing various basic human movements such as standing, sitting, squatting, lying down, crawling, etc., and can also customize body shape and gender according to user needs. In addition, the robot is equipped with a head with 3D curved surface display, integrated with a large number of sensors, capable of expressing emotional states, and equipped with dexterous hands and advanced technologies such as all-solid-state batteries to achieve safer and more flexible operation. Despite facing many technical challenges, such as the high number of 22 degrees of freedom in the hand joints, the Xiaopeng team is dedicated to creating a robot partner with high intelligence, ease of use, and personalized features.

A：The all-new generation of Elantra is equipped with three Turing AI chips, with a computing power of 2250 tops, making it the highest computing power integrated in the world.

A：Elon, based on powerful computing power, has loaded large-scale models of the physical world, including VLT, VOA, and VLM. In particular, Xiaopeng has pioneered a large-scale model called VLT, which is the core engine for robot autonomous action, responsible for determining how the robot should act next in a real physical environment. Xiaopeng Motors has established the first physically intelligent data factory, dedicated to acquiring large amounts of data, even long-term massive data, to promote the long-term high-speed generalization of robots. In addition, they are expanding the three laws of robotics, considering adding a fourth law to protect human privacy data, as well as an active safety protection law to ensure that robots will not harm humans unintentionally in any situation.

A：The movement model of robots is much more complex than cars, as they have 82 joints and can create a wide variety of motion combinations. Precise control is needed from the feet to hands, waist, shoulders, and neck.

A：Xiaopeng has deeply considered the commercial mass production, and believes that robots entering factories to screw or do housework is not the most optimal scenario. Currently, Xiaopeng has chosen guidance, shopping guide, and search guidance as the first landing directions in commercial scenarios, and plans to train robots to understand natural language in order to achieve operation without a remote control.

A：The challenges facing Xiaopeng include the need for hardware design to adapt to data-driven models, high integration innovation between software and hardware, and meeting the quality requirements of vehicle regulations or even higher. To address these issues, Xiaopeng has set up a large-scale R&D team of over 1000 people, integrating the strength of ten R&D teams and nine collaborative departments, committed to self-developed full-stack software and hardware integration innovations. The goal is to achieve mass production of advanced humanoid machines by the end of 2026.

A：We believe that the ultimate value of AI lies not only in technological breakthroughs, but also in becoming an extension and partner of humanity in the physical world. It is not about replacement but about empowerment, not about subversion but about symbiosis. Artificial intelligence is the core driving force behind a new wave of technological revolution and industrial transformation, as well as a strategic high ground in the global competition of advanced manufacturing.

A：The all-new generation of Xiaopeng Motors has the most human-like hardware form, and will achieve mass production of all-solid-state batteries in 2026. We hope to inject new impetus into the high-quality development of China's new energy vehicles and the intelligent transformation of the manufacturing industry through deepening cooperation.

A：Xiaopeng has made significant progress in the field of low-altitude aviation, such as the development of the flying car A868, which is more efficient than cars and more flexible than high-speed trains. It has aviation-grade performance, can achieve vertical take-off and landing, and is expected to go into mass production in 2026. Meanwhile, Xiaopeng Huitianzheng is promoting the development of low-altitude economy in two directions: one is the land-based aircraft carrier-type flying vehicle, and the other is a flying car that can take off and land at specific locations with long endurance. The flying car A868 strictly adheres to the highest flight safety standards, using a six-axis and six-propeller configuration with electric ducted fans for safety, ensuring safe flight and landing even if a single engine fails. Xiaopeng executives have personally experienced flights of over 5000 kilometers to ensure safety, and plan to launch flying license training services.

A：Xiaopeng showcased a number of innovative achievements at the Technology Day, including low-altitude flying cars, next-generation robot technology, the application of AI in physical world modeling, and the upcoming high-performance humanoid robot. In addition, Xiaopeng will also showcase the intelligent manufacturing factory at the new headquarters in Guangzhou, and will hold the X9 Kunpeng Super Extended Range Technology Conference on November 6th, announcing the start of a new era with the adoption of dual-energy sources.

A：Today Xiaopeng will release four important AI-based technology applications, representing Xiaopeng's new breakthroughs and advancements in autonomous driving, intelligent hardware, and the integration of software and hardware. In the past six Tech Days, Xiaopeng has proposed plans for high-speed autonomous driving for commercial use, nationwide popularization of autonomous driving, flying car concepts, robots will truly land in the coming years, applying autonomous driving technology to $20,000 cars, and next-generation supercharging, among many controversial and visionary technological ideas, and these ideas are gradually becoming reality.

A：Xiaopeng made significant exploration and innovation on the VLA autonomous driving operating system, from the initial software writing rules to the current exploration of directly translating visual information into actions in the physical world model, especially the research and development of the second generation VIA (physical world model), which achieved a close integration of models, computing power, and data, significantly improving the performance and safety of the autonomous driving system, and achieving a new paradigm of mass production of physical world models two years ahead of schedule.

A：The theme of "emergence" originates from the AI large model field, and in the field of Xiaopeng's autonomous driving, it manifests as a new combination and discovery of data and capabilities after training with a large amount of data. For example, the new products, new scenarios, and new capabilities exhibited by Xiaopeng in cross-domain integrated new products such as flying cars and humanoid robots are specific manifestations of emergence.

A：Xiaopeng believes that accurately grasping long-term trends is crucial for the development of enterprises. In the digital world, operating systems are transitioning from rule-driven to data- and model-driven, driving changes in intelligent hardware and platform ecosystems. In the physical world, the automotive market is undergoing a major transformation from fuel-powered cars to electric cars. The deep integration of AI with the physical world will bring about huge disruptive changes, ushering in a new era of physical AI.

A：When developing the second generation VI, Xiaopeng experienced months of failures and faced immense financial pressure. However, in the end, he achieved a stunning emergent effect through a physics world model transfer based on a low latency principle. This model enabled Xiaopeng to unexpectedly solve many problems that were previously unsolvable while addressing a single issue.

A：The second generation VI is centered around vision, learning and observing the world like a human, without the need for intermediate language translation, directly converting it into motion trajectories. This model significantly reduces information loss, increases reasoning efficiency, improves frame rates, and provides a safer and smoother driving experience. It is trained using data from nearly 100 million video clips, with a special focus on corner cases and long-tail data, in order to achieve a more comprehensive understanding and predictive capability of the world. Additionally, through the optimization of the entire chain of chips, operators, and models provided by Alibaba Cloud's ultra-powerful computing cluster, a powerful physical world model has been built that can simulate real environments and make decisions.

A：The second generation of VI will be launched in the NP lane park, enhancing urban assisted driving capabilities, especially performing well in complex small roads and park scenarios, taking over the average mileage by 13 times, making the experience from parking space to parking space smoother. This technology is not limited to China, but can also adapt to various road conditions globally, enabling Xiaopeng to take a big step forward in the global autonomous driving field.

A：Yes, Xiaopeng is currently developing the technology of the second generation VA from chip to EEA architecture, model, and even partners, and is open to collaboration with global vehicle manufacturers and suppliers to promote the technology from China to the global market, expanding the Xiaopeng brand to more car models.

A：Volkswagen will become the first customer of Xiaopeng's second generation VPA, and Xiaopeng's Turing AI chip has been designated by Volkswagen. Building on their cooperation over the past few years, both parties are jointly driving the advancement of autonomous driving technology in China and globally.

A：Xiaopeng's newly developed VOS can achieve efficient voice-controlled driving on local internet, supporting mixed instructions in multiple languages, and can provide low-latency autonomous driving interactions based on user voice commands, whether in a parking lot, on the road, or in the wilderness, it can provide a seamless voice control experience.

A：When using Xiaopeng Motors abroad, I cannot use voice commands to wake up the vehicle for navigation and conversation in the UK, France and Germany due to language barriers, as I do not understand the languages in these countries.

A：Starting from 2026, Xpeng Motors will not only support the Android operating system, but also will introduce a new language interactive dialogue memory operating system called VIM. This system is currently under development, and it is expected to bring significant changes and differences in the Xpeng Motors based on VIM next year and the year after.

A：Xiaopeng's second-generation VA large model is a new paradigm of physical model, based on the improved basic model of VA, with new capabilities and emerging changes. This set of models can be used not only in Xiaopeng cars, but also in Xiaopeng's robot taxis, flying cars, and will be combined with VIM to bring new capabilities and experiences. Xiaopeng plans to launch three unmanned vehicles in 2026, using a full-stack self-developed hardware and software solution to solve the limitations faced by autonomous driving vehicles, and is committed to scaling and commercializing robot taxis from China to the world. At the same time, Xiaopeng will also start unmanned driving trial operations in cities such as Guangzhou.

A：The first type is a driverless car, known as a robot taxi, which operates under a fully shared model. The second type is a car that individuals can purchase and own with Level 4 hardware and software experience, but can be driven by professional drivers when needed, or used for shared familial use within the household. This model complies with current legal requirements and can be considered as a Level 2 car.

A：The reasons for the difficulty in popularizing autonomous vehicles include high costs, complex legal and regulatory requirements leading to a limited number of fleets, poor profitability in commercial operations, and limitations in autonomous driving technology, such as the inability to operate in areas without high-precision maps, complex small roads, or nighttime driving scenes.

A：Xiaopeng's unmanned vehicle design is oriented towards Level 4 intelligent driving standards, equipped with a full set of hardware redundancy to ensure driving safety; it has a computing power of up to 3000 pps, and has trained automatic driving strategies based on the second generation VIM that prioritize safety and full hands-off driving experience. In addition, it has also innovated with an external vehicle interaction system, such as VLM and external voice functions, allowing the vehicle to communicate and interact with pedestrians. Xiaopeng not only plans to put unmanned vehicles into commercial operation, but also considers applying related technologies and products to Level 4 experience vehicles for ordinary consumers, exploring the market where unmanned vehicles coexist with human-driven vehicles. Xiaopeng's unmanned vehicles will have full safety redundancy, powerful computing power, multiple self-driving modes, and may also try out more innovative service models, such as cloud-based security officer services.

A：Xiaopeng announced a global ecological partnership with Gaode Maps to jointly promote the extensive application and development of Xiaopeng robot taxis in China and even globally. Both sides will integrate resources and innovate technology to create a more intelligent, efficient, and seamless travel experience, and jointly shape the future of intelligent transportation ecology.

A：In a household environment, the main problem faced by quadruped robots is their inability to turn smoothly, especially in narrow spaces such as near a bedside table. Turning in place can lead to collisions and damage. Unlike soft-bodied animals like dogs and cats, the rigid structure of quadruped robots can cause significant issues when turning in a home environment. The choice of humanoid robots is because their structure allows valuable data to be easily obtained from human bodies for training. Additionally, most living scenarios are designed for humans, making humanoid robots easier to generalize to different environments and more readily accepted commercially, aligning with users' aesthetics and preferences.

A：The new generation of Alpha robots has entered the mass production preparation stage, and it is expected to achieve the mass production of hardware and software integration around April next year. The prototype currently being displayed can achieve a light and graceful walking posture, with a spine structure similar to that of humans, capable of performing basic actions such as bending over and touching the ground with both hands. It also has bionic muscles and soft skin, with a flexible and customizable body shape. The new generation of robots has a human-like skeleton, muscles, and skin, capable of performing various human basic actions, supporting body type customization, and featuring 3D curved surface display on the face and a large number of sensors to express different emotional states. In addition, it is equipped with industry-first solid-state batteries, ensuring safety and high computing power with a Turing AI chip, and is equipped with VLT large models as the core engine for the robot's autonomous actions to achieve more intelligent behavioral decision-making and motion control.

A：Xiaopeng has added a fourth rule in the field of robotics ethics - not to disclose people's private data, and has established a proactive safety protection rule to ensure that robots will not inadvertently harm people, animals, or other objects under any circumstances. At the same time, in terms of commercial applications, Xiaopeng has found through in-depth research that humanoid robots may be more suitable for home scenarios in the Chinese market compared to scenarios such as screwing in factories, but the specific commercialization process is still being explored.

A：Currently, there is no robot company in the world that can guarantee absolute safety in a household environment with less than 100 square meters and lots of clutter, ensuring that no accidents occur. Therefore, safety is the biggest challenge for robots entering households.

A：Current technology and capabilities are still not sufficient for robots to understand and adapt to various household environments. Therefore, in the short term, it is difficult for robots to achieve generalized learning in a household environment. Xiaopeng chooses to first practice in three directions on the islands, namely tour guidance, shopping guidance, and search guidance (front desk guidance). For example, after simple training with natural language, robots are guided to perform shopping guidance in exhibition halls, sales stores, and gradually more complex tour services. However, test driving or search services are not yet possible.

A：Xiaopeng faces a huge challenge of integrating innovation when mass-producing robots. Unlike the integration and innovation model in the car industry, robots need full integration of software and hardware to achieve full-stack independent research and innovation with hand-eye-foot coordination. Xiaopeng has established multiple research and development centers and collaborative departments to jointly develop integrated robots, with the goal of achieving mass production of high-end humanoid robots by the end of 2026.

A：Xiaopeng's robot is different from traditional cars, as it is driven by software rather than hardware. For example, in hardware design, one must consider the impact of data-driven models, software, and AI on the hardware, and make changes to the architecture. In addition, Xiaopeng's requirements for the quality of the robot are even higher than that of a car, such as using all solid-state batteries to avoid the risk of fire, and having very high safety requirements for joints, body, and other components, even exceeding automotive standards.

A：In the future, there may be new positions specifically responsible for training, teaching, and managing integrated robots. Xiaopeng is considering mass production of robots, expecting to reach the ability to mass produce sample A by the end of the first quarter or second quarter of next year. At that time, the robots will be able to interact and move with users, and will have a certain level of security, although they may be slow but efficient.

A：Yes, Xiaopeng's robot will be open to global developers like the second generation VA large model and robot taxi, encouraging developers to participate in secondary development to enhance the application scenarios and capabilities of the robot, creating infinite possibilities.

A：We have already launched the first low-altitude self-driving tourism route in northwest Dunhuang, and plan to build five flight camps around March 2023, so that everyone can explore Dunhuang in a new way and overlook the breathtaking scenery. At the same time, Huitian has established the country's first low-altitude flight camp, landing in Guangzhou University City, and is about to move to Huitian's new location, where the country's first 6S flagship store will be opened, integrating display, sales, delivery, and training services.

A：The Huitian flying car will launch an exclusive one-on-one training for flying licenses, which means in the future people will not only need a driver's license, but also a flying license. Our designed land carrier type flying car can be operated with just one screen and one control lever, allowing for single-handed operation. Even if the main control lever fails, there is a backup control panel to ensure a safe landing. Welcome to experience flying firsthand next year.

A：In the park next to Xiaopeng Campus, we will soon open a flying experience. Currently, the first production version of the flying car has officially rolled off the assembly line in ET0 state, and is expected to start delivering to China and the world in the second half of next year, providing a brand new flying experience. Hovering Flying Car is expected to set a new global sales record for manned flying vehicles after mass production and delivery. We are committed to promoting flying cars to more regions globally.

A：On Technology Day, we introduced the development progress of the second generation model of Xiaopeng VAA. This is the first physical world model that can be put into mass production, which will accelerate the arrival of autonomous driving. At the same time, we plan to launch the robot tec and a higher-level intelligent configuration noble version in 2026, and are committed to becoming the first Chinese enterprise to mass produce high-level humanoid robots.

A：Xiaopeng has demonstrated the advantages of its self-developed full-stack software and hardware from dream to the process of coming soon to landing in the low-altitude field, and looks forward to becoming an explorer of the original intention in the world of physical AI, committed to building a global intelligent travel solution provider. In the future, Xiaopeng's new headquarters and intelligent manufacturing factory will be open to visitors, showcasing a variety of cutting-edge technologies including flying cars.