Jidu Auto—the intelligent EV (electric vehicle) startup founded by Baidu Group, backed by Zhejiang Geely Holding Group, and based in Beijing—has unveiled its first “concept production robocar” called ROBO-01. The concept car had been slated for a reveal at the 2022 Beijing Auto Show this April, but the COVID lockdowns there got in the way. However, it was unveiled at the startup’s first Roboday held in the “XiRang metaverse.” At the unveiling, a “digital human car owner” named Xijiajia drove and interacted with the concept.

Having transitioned from the “Fuel Car 1.0 Era” to the “Electric Car 2.0 Era,” Jidu says that the automotive industry is entering the “Intelligent Car 3.0 Era”—and it is leading in the emergence of a new generation of automotive products. With its futuristic design, the company says that its robocar is set to revolutionize the automobile industry, leading the way to an intelligent car era driven by artificial intelligence (AI) created by Baidu.

“The Intelligent Car 3.0 Era is the era of robocars,” said Xia Yiping, CEO of Jidu. “The transition to this new era is marked by the shift of driving power from humans to AI… The automotive industry in the 3.0 Era will see a seismic shift from a revolution in energy to a revolution in product attributes. The ultimate goal is to realize a fully driverless transportation experience. The Jidu robocar aims to meet users’ needs for intelligent travel, in-car intelligent assistance, and intelligent cabin in the new era.”

The concept’s capabilities are based on unique “trainable” functionality and neural JET (Jidu Evolving Technology), which support a high-level autonomous driving solution with full redundancy, service-oriented architecture (SOA)-based intelligent cabin, and millisecond-level offline intelligent voice assistant. Other highlights include an active, deformable body-structure design; a 3D borderless one-piece large screen and zero-gravity seating inside; AI pixel lights; and dual max-computing chips.

 

High-tech backing

Developments have been moving quickly for the new OEM.

In January 2021, Baidu announced its plan to enter the automotive industry by launching a smart car company, and two months later Jidu was established, receiving over $300 million in startup capital. About a year after announcing the launch, the new smart car company had closed its Series A financing, raising nearly $400 million. That round of financing, backed by Baidu and Geely, meant that the startup could accelerate its R&D to achieve production sooner.

The R&D expansion will focus on talent acquisition, especially for advanced autonomous driving, smart cockpit, and smart manufacturing. The company’s goals with the robocar are an outstanding user experience and cost-effectiveness that set industry benchmarks for smart cars. Helping in that effort are global suppliers the company highlights on its website: Aptiv, Bosch, CAIP, CATL, Continental, Grammar, Hella, Hesai, Magna, Nvidia, Qualcomm, Renesas, Suspa, UAES, and ZF.

The company plans to officially launch a limited-run version of its first production model in the fall. It will be 90% similar to the ROBO-01 concept car. It will also unveil the design of its second production model at this year’s Guangzhou Auto Show in November.

“With the support of cutting-edge AI technology combined with a high-quality vehicle platform and manufacturing process, Jidu’s efficient development of [an] automotive robot has been proven; the brain, nerve system, and body of the robocar are all under rapid development,” said Xia. “Since our formation last year, Jidu has begun mass production software R&D verification based on the software-integrated simulation prototype SIMUCar, bringing innovation and high efficiency in the field of intelligent development. By the end of 2021, SIMUCar successfully demonstrated intelligent assisted driving in both urban and high-speed areas.”

 

Exterior and interior design

The ROBO-01 features a simple and futuristic look and four center-opening doors. The front butterfly wing doors open out and up, with the rear pair of doors swinging back conventionally. Its non-marking side windows are said to be the first of their kind in the industry. An active rear wing aids aerodynamics.

The concept is equipped with a set of fully adjustable structures including front-hood-mounted “collapsible” lidar sensors. They are said to enhance the safety and stability of the intelligent driving system and strengthen the sensing capabilities for advanced autonomous driving. The lidar can be collapsed by AI intervention before a crash occurs, with the startup obtaining a Chinese national patent for the technology.

The robocar can recognize the user’s emotions and interact with the outside world by expressing its own “emotions.” Its front design integrates interactive AI pixel headlights and an exterior high-recognition-rate AI voice interaction system for natural communication between humans, vehicles, and the environment.

Inside, adjustability is a hallmark of the foldable U-shaped steering wheel, liftable satellite speakers, and adaptive zero-gravity seats.

The U-shaped steering wheel is meant to maximize visibility of the large display screen. In addition, steer-by-wire technology supports the steering wheel’s folding/hiding movement and enhances the variable steering ratio in automatic driving mode.

The spacecraft-inspired zero-gravity seats are lightweight and breathable, wrapping around the occupant and including a “swan neck” headrest design with an adaptive adjustment function.

The main, 3D, borderless, ultra-clear display uses a non-spliced screen running from A-pillar to A-pillar and is a key element in a promised immersive audio and visual experience. The door handles, shift lever, left and right indicator levers, and other physical control keys have been removed, which is said to make the human-vehicle interaction experience even more natural and seamless.

The intelligent cabin’s off-line voice assistant does not rely on network signals so it can provide a “millisecond-level” response. The full off-line intelligent voice function. Its multi-modal “human-like” interaction capabilities—such as visual perception and lip capture—are said to allow for “natural and smooth” communication between user and car. Other HMI highlights are a 3D human-machine co-driving map and “full-scene interaction” inside and outside the car.

The company was the first to announce a car to launch with the fourth-generation Qualcomm Snapdragon automotive cockpit platform. It enables the 3D presentation of the ultra-clear large screen for navigation, games, entertainment, online office, and other scenarios.

 

Intelligent and autonomous driving

With the concept, Jidu says it is the only smart automaker in the industry that applies Baidu’s full-stack Apollo autonomous driving safety system. The system’s capabilities have been extensively used in Baidu’s robotaxi, with 27 million km (17 million mi) of safe autonomous driving. The large robotaxi road test fleet is being road-tested in more than 30 cities across China, which is continuously improving the system’s ability to cope with complex urban roads.

Baidu says that it has 13,000 AI patents, ranking first in the number of AI patent applications, authorizations, and patent quality in China. It also claims to be the world’s first in the field of deep-learning applications, with 3400 intelligent driving patents.

The Jidu intelligent driving system is powered by Baidu’s autonomous driving computing platform using Nvidia Drive Orin, while its software and algorithms are jointly developed by Jidu and Baidu. The system features “dual” Orin X chips and 31 sensors including 2 lidars, 5 millimeter-wave radars, 12 ultrasonics, and 12 cameras.

Orin, which is designed specifically for software-defined cars and is reaching the market in production vehicles later this year, enables continuous upgrades throughout the car’s life via over-the-air software updates. It processes 254 TOPS (trillion operations per second) of computing performance and can support autonomous driving functions, confidence view visualization, digital clusters, infotainment, and passenger interaction AI.

Based on Jidu’s self-developed SOA, the intelligent driving system is said to be an industry-leading “true redundancy” dual-system solution. It has been successfully tested on the SIMUCar “software integrated simulation vehicle,” said to have verified the safety and stability of the advanced autonomous driving system for mass production.

Capable of point-to-point advanced autonomous driving, the system can adapt to three main driving scenarios: high-speed, urban roads, and parking. The system has been tested and verified to handle unprotected left turns, traffic light recognition, obstacle avoidance, and freeway on/off ramps. Users will be able to access these and other advanced autonomous driving functions from the moment they begin driving.

“The Jidu vehicle has been designed to operate under the concepts of free movement, natural communication, and self-improvement,” explained Robin Li, Baidu Co-founder and CEO, at Baidu’s annual flagship developer’s conference Baidu Create 2021. “First, the Jidu self-driving vehicle features L4 autonomous driving capability for free movement. Second, human-car interaction and accurate identification of speech semantics enable the car to communicate naturally. Third, the vehicle can self-learn and iterate based on user’s habits and constantly optimize autonomous driving features and smart cockpit experiences to realize self-growth.”

Free movement refers to the L4 autonomous driving capability supported by Jidu’s intelligent driving system, which is said to cover reliable autonomous driving functions across high-speed and urban open road environments.