MOUNTAIN VIEW, CA --- CEVA, Inc. and AdasWorks will be showcasing a low-power ADAS solution demonstrating Free Space Detection at the upcoming AutoSens Conference in Brussels, Belgium, September 20-22, 2016 -- Booth #21.
The solution combines AdasWorks’ state of the art convolutional neural networks (CNNs) with the CEVA-XM4 intelligent vision DSP, to deliver a cost- and power-optimized system capable of detecting the free space surrounding the vehicle to provide a safe path for autonomous vehicles. The CEVA Deep Neural Network Software Framework (CDNN2) was used to convert AdasWorks’ pre-trained networks to run on the CEVA-XM4, significantly simplifying the integration and optimization process. Using CEVA’s CDNN2 drastically improves the time to market when moving from networks developed on platforms tailored for research and development into low power, cost efficient platforms targeted for high volume applications.
AdasWorks develops Drive 2.0, an AI based full stack software suit for Level 5 self-driving cars. Drive 2.0 will provide the full spectrum of automated driving functionalities from environment recognition, localization and path planning to low-level control. The Budapest-based company is also designing Toolkit 2.0, a framework combining the training and testing tools essential to build the Drive 2.0 suite. The toolkit will offer 360-degrees solution for the building of safe self-driving vehicles from data collection to real-time simulation and testing. The company has over a 100 AI, computer vision, navigation and automotive engineers working on the development of a hardware-agnostic and scalable software solution for drivers all around the world. AdasWorks is set to unveil their own self-driving car using affordable cameras as primary sensors at CES 2017 in Las Vegas. The company opened its US office in Mountain View this September and will open offices in Asia and Northern Europe in the near future.
The CEVA-XM4 intelligent vision DSP efficiently addresses the intensive processing requirements of the most sophisticated computational photography and computer vision applications such as video analytics, augmented reality and advanced driver assistance systems (ADAS). By offloading these performance-intensive tasks from the CPUs and GPUs, the highly-efficient DSP dramatically reduces the power consumption of the overall system, while providing complete flexibility. The platform includes a vector processor developed specifically to deal with the complexities of such applications and an extensive Application Development Kit (ADK) to enable easy development environment. For embedded systems targeting deep learning, the CEVA Deep Neural Network (CDNN2) real-time neural network software framework streamlines machine-learning deployment at a fraction of the power consumption of the leading GPU-based systems. For developers of Advanced Driver Assistance Systems (ADAS), an ISO 26262 compliant safety design package is available for the CEVA-XM4 that helps customers accelerate their certification.