BMW was selected by Nvidia for updating of a system of logistics

2020: Updating of a system of logistics

On May 14, 2020 the BMW Group company reported that it selected the platform of creation of NVIDIA Isaac robots for upgrade of the automobile plants and creation of logistic robots who will use technologies of AI calculations and visualization from NVIDIA.

According to the company, cooperation assumes use of NVIDIA technologies at all stages of creation of an intelligent logistics system: from training and testing before its deployment. BMW robots will be developed on the basis of program architecture which is used on the open NVIDIA Isaac platform for creation of robots. Such system will be installed at the BMW Group plants worldwide.

Production of the qualitative, customized cars in bigger volume on one production line requires the advanced computing solutions. Jürgen Maidl, the senior vice president for logistics in BMW Group told

Within cooperation of the companies capacities supercomputers for AI NVIDIA DGX and Isaac technology for training and testing of robots are used; processors ray tracings, graphic with support, NVIDIA Quadro for rendering of synthetic spare parts for training improvement; and also a line of the AI robots constructed on SDK Isaac managed by edge systems NVIDIA Jetson and EGX.

Independent logistic AI robots who participate in assembly of cars with necessary extent of customization on the same production lines help to optimize logistics.

BMW Group is going to apply the NVIDIA Isaac platform to development of five types of AI robots for improvement of logistics based on different edge of NVIDIA Jetson AGX Xavier and EGX computers. Also navigation robots for autonomous transportation of materials, and manipulators for selection and sorting of parts treat them. The robots developed on NVIDIA Isaac SDK will be controlled the deep neuronets which are responsible for perception, segmentation and assessment of a position of the person for recognition of the surrounding among and objects and autonomous navigation and movement of objects. These robots are trained both for real, and at synthetic data sets using the graphic processors NVIDIA for rendering of spare parts by a ray tracing method in different conditions of lighting and noises for addition of real data.

Further real and synthetic data are used for training of neuronets at the NVIDIA DGX systems. Then robots take tests for navigation and manipulation in NVIDIA Isaac simulators based on the NVIDIA Omniverse platform where specialists of BMW Group from different regions can work jointly in unified virtual environment.