As more and more autonomous driving cars go out from labs to street and serve the public it comes to a critical moment that autonomous driving technology approaches to the phase of mass productivity. To better prepare the upcoming first year of autonomous-driving commercialization ear in 2020, currently almost all the entrants are accelerating the pace of research and development, in order to promote the product to market.
September 17, In-Driving, startup focusing on connected autonomous driving systems and components, launched their robust autonomous driving software suite named Athena in Shanghai. Athena can runs on In-Driving’s embedded domain controller named TITAN smoothly and robustly, which means In-Driving has finished the layout on autonomous driving with integrated software and hardware.
In-Driving had released its newly domain controller TITAN III in Jun. 8th this year. On the board of TITAN III, four Nvidia Jeston TX2 modules are integrated to support computing force for autonomous driving algorithms. The board runs UBUNTU operating system, and supports multiple hardware interface accesses such as USB, CAN, RS232, Ethernet, AD/DA, etc. Besides, the controller can be connected to maximum 12 cameras, 6 Lidars and 5 radars at the same time. Ublox modules m8p and m8l are also integrated to provide basic position information. What’s more difference, V2X and 4G communication chips are also integrated to provide connected ability. High integration of TITAN III can support all the needs of integrates all L4 self-driving functions in one board. By the enable of Athena, multi-sensor fusion of 3D perception, high precision positioning, vehicle control, interactive communication with cloud can be installed without any other external devices. You only needs to connect the cameras, lidars and radars to realize a variety of intelligent functions such as high precision map, ACC, LDW, LKA, FCW, LC, VUI, PDS, AEB, one-click car calling and drop-off and so on.
By combining Athena with TITAN autonomous-driving domain controller, IN-DRIVING realizes the HD map collection, real-time 3D fused perception and localization, data storage and high-speed Ethernet communication, as well as diverse autonomous driving by single car, platoon and car to infrastructure communication. The flexible solutions make it possible to shorten R&D and testing time and fasten the progress of mass production.
In the product launch, the main leaders of In-Driving introduced their critical progresses on how using current AI to solve self-driving tasks, including 3D perception, multi-layer localization by multi-sensor fusion, reinforcement learning for motion planning and decision etc.. They also demonstrated the one-click car calling and drop-off, and showed the live of 5G connected autonomous driving in Beijing.
Dr. Jing Cao, director of In-Driving Corporate Research, said that some new changes have been taken place in the autonomous driving area by means of AI’s new development, such as the change from 2D to 3D perception, including the physical description of the three-dimensional information, as well as direct deep learning by point cloud.
Prof. Jianwei Zhang, co-founder of In-Driving and Academician Hamburg of Germany, gave a talk on the next challenge of artificial intelligence and the applications. He pointed out that the current AI was weak AI, which was easy to make big mistakes instead of small mistakes. He’s research focused on how to use multi-modal information of humanoid to achieve strong AI, which is one of the most important basic research topics. This topic is also founded by Sino-Germany cooperation project and hosted by Prof. Jianwei Zhang. He also pointed out that we can use and reference current achievements in reinforcement learning, continuous learning and transfer learning to improve the learning ability of automatic vehicles. Zhang also mentioned the need to pay more attention on ethics of autonomous driving, especially morality.
Besides, Dr. Ming Li, CEO of In-Driving, said the solution of software and hardware integration of was nearly completed, and the next phase of the company would focus on how to further improve the reliability of the open platform and make this solution commercially available as soon as possible.
“In November, at Nvidia’s 2018 China GTC Conference, we will release specific commercial plans, including In-Driving’s autonomous-driving solutions in the business scenarios of heavy trucks for trunk logistics and passenger vehicles for travel service.”