Chinese car maker NIO has launched a selfdriving car that it says can be used in motorway and urban environments when it hits the road in 2022.
The ET7 electric sedan, detailed at the weekend at the company’s annual event, uses a range of 33 sensors including 11 8Mpixel cameras that can detect another car at a distance of 225m, 1550nm lidar with a 500m range, five mmWave radars and 16 ultrasound sensors for local sensing, including autonomous parking. The lidar system has dynamic focussing on areas of interest, for example at a crossroad to focus on vehicles and pedestrians in urban environments.
Like fellow Chinese manufacturer Xpeng, NIO is looking to take on US electric car maker Tesla with its FSD full selfdriving technology.
In the NIO ET7 the sensor data is combined in a board with four Orin processor from Nvidia that deliver over 1000 TeraOperations/s. This liquid cooled ADAM system has 48 ARM Cortex-A78 processor cores, 256 Tensor AI cores and 8096 CUDA GPU cores with 1016TOPS peak performance to run the NIO Autonomous Driving (NAD) full stack software. The system has a 6.4GPixel/s wide I/O for the data and the four chips are connected by a 32GB/s backbone network. Two of the Orin chips are used for control, with the third for redundancy and the fourth allowing the AI neural network to evolve with additional learning data
“This is by far the most powerful mobile computing platform on a production model,” said William Li, CEO of NIO.
Inside the vehicle there is a head up display, 12.8in AM OLED 5mm thick control display and 10in instrument cluster powered by Qualcomm’s Snapdragon platform with WiFi6, ultrawideband (UWB) for the key fob and 5G cellular and V2x wireless links to smart infrastructure and other vehicles.
The NAD system will be available on subscription for RMB680 ($105, €85) a month and NIO is taking pre-orders now.