Our CEO Maxwell Zhou took the stage at the 26th Guangdong-HongKong-Macau GBA auto show today to share his thoughts on the commercialization and scalability of L4 solutions. Here's the highlight of his speech.

Production Ready and Affordable

We launched Driver 2.0 in December 2021, the first to make L4 solution under $10,000 and meet autograde requirements. In April, we integrated Driver 2.0 into the latest Robotaxi fleet and its cost can be reduced by 70% for mass production. Driver 2.0, composed of 8 cameras, two autograde Orin chips, 2 to 5 solid-state LiDARs and GNSS, can be mass produced and customized for the clients' needs.

Our Driver 2.0 has been validated in real-life scenarios across different cities in CBD areas. Whether it's heavy rainfall, snow, or on congested roads in morning rush hour or highway, our Robotaxis performed smoothly and smartly.

3 Prerequisites for Mass Production of L4 Autonomous Driving

To achieve mass production of autonomous vehicles, we think there are 3 prerequisites: auto-grade sensors, chips and HD maps, all of which are powered by algorithm and toolchain. Sensor price has seen significant decreases in recent years. As they have met auto grade requirements, solid-state LiDARs are installed on many vehicles. For chips, we integrated cutting edge NVIDIA DRIVE Orin SoCs engineered to fulfill safety standards and simultaneously process various applications critical for AV decision-making. HD maps are also indispensable in terms of large scale deployment of autonomous vehicles. Lane-level accuracy covering vast areas and continuous updates are required to guarantee the safety of autonomous driving. We have been adapting our algorithm to the HD map partner since last year.

Our strategic advancements on these 3 fronts lie in our competitive edge of algorithms. At DeepRoute.ai, our founding team gave a lot of deliberation on the end game of autonomous driving, which is to popularize Robotaxi and L4 consumer vehicles so individuals can enjoy the conveniences brought by this technology. Guided by this strategic thinking, we advocated early fusion in 2017 and developed inference engine in 2019, both of which are the first in AV industry. Planning and control algorithms are based on game theory to solve various vehicle-surroundings interaction scenarios.

We consistently think before L4 autonomous driving can be commercialized and scaled up, efficient algorithms, low cost hardware and low power consumption need to be achieved.

3 Years We Turn Sci-fi into a Reality, and a Better One

We are honored to be the first to operate in CBD area of Shenzhen and by May 27, 2022, we have operated 313 days with over 40,000 orders, and over 96% gave us 5-star rating. Many citizens were impressed that what used to exist in science fictions is now a living reality.

Apart from daily operations in the city centre carrying people around for work and leisure, we are also committed to bringing autonomous driving to the betterment of society at large. Robotaxis were deployed immediately to facilitate transportation of Covid-19 test samples to the local hospital in July 2021, and our autonomous driving trucks' first public appearance was on the frontline of restricted zones delivering medical goods and food to local communities. Our ride-hailing platform is also adapted to visually-challenged groups, and Robotaxis were installed with voice assistance to help them locate the vehicle and the in-cabin help button is ready to help whenever needed. We will continuously devote morse efforts to making future mobility a faster reality, and more accessible.