The advanced driving experience need the support of hardware and also the software. NVIDIA announced that DRIVE software has been updated to version 8.0, and have more new AI based functions. Such as enhanced perception, visualization, map planning and more. They are expected to release DRIVE software 8.0 for developers in middle of January.
DRIVE software is an open source software package and is powered by DRIVE AGX. It include the DRIVE OS operating system, DriveWorks software framework and the SDK self-driving programs, also the DRIVE AV and DRIVE IX.
Developers can combine high resolution visualization technology with in-vehicle sensors with this software update. With the car visualizaion, they can provide self-driving instant vision for driving, showing the real time visual information, so that they can plan the subsequent driving route immediately. In addition, DRIVE Softwear also updates and improves object detection capabilities and provides 360-degree panoramic images of vehicles.
DRIVE IX can track the input informations and sensing sensing information of the self-driving vehicles. The developer can read the driving data by the real-time information and previously stored informations in the database.
Vehicle information visualization can be divided into five stages, including sensing, perception, map, planning and driving. First, the screen will detect the unprocessed camera and sensor data and decide the object's position.
The bounding frame is an object detection program, and at the same time, the object is marked and displayed from the object at a distance of up to 50 km, and the display screen can mark the lane line perceived by the self-driving vehicle, and the following shows that the vehicle will travel by the planning stage. The route finally shows the direction of the self-driving, acceleration and braking, and provides easy-to-understand visual information during the driving phase.
In addition, DRIVE IX also provides driving monitoring. At the same time, it also has a stage similar to the automatic driving visualization. It will drive the camera image information of the driving and analyze the driving through the deep neural network to judge whether the spirit is distracted or distracted by the face and eye tracking. Visually remind the driver and the passenger of the process of real-time analysis, while displaying the status in a red tone that may indicate a dangerous situation with a warning.
Autopilot 2.0 is artificial intelligence and Tesla HUD! | Model 3 Owners Club
DRIVE AV provides more sophisticated machine vision analysis and safer planning of route. DRIVE Softwear 8.0 also brings more advanced machine vision recognition to DRIVE AV. LightNet not only recognizes the main and directional lights of the vehicle, but also identifies green, yellow and red lights, while SignNet has the identification of traffic signs. Ability to distinguish speed limit, one-way street, no-pass and parking re-opening information.
In addition, the new version of the software DRIVE AV also added DRIVE MapStream, enabling developers to upload car sensor information to their digital maps, so that self-driving vehicles can confirm the location on the road and predict whether there are obstacles such as construction in front of the road.