[QNX]QNX Platform for ADAS 2.0

鋒影

email:[email protected]

如果你認爲本系列文章對你有所幫助，請大家有錢的捧個錢場，點擊此處贊助，贊助額1元起步，多少隨意

Autonomous cars contain the most complex hardware and software everdeployed by automakers. The software must process a flood of data from sensorssuch as cameras, LiDAR and radar in real time to form a model of the car’ssurroundings and make safe decisions on control of the vehicle. This requireshighly ecient, safe and secure software that can make use of special purposehardware (accelerators) for vision processing and deep neural-net basedmachine learning algorithms. The QNX Platform for ADAS provides a functionallysafe software foundation that enables the above functions to deliver anadvanced driver assistance system (ADAS) for automated driving.The QNX Platform for ADAS is built upon the QNX OS for Safety, which iscertified by TÜV Rheinland to ISO 26262 ASIL D. This certification also includestool chain qualification for the C compiler, assembler and linker at level TCL 3.

Functional safetyISO 26262, an international standard for functional safety, is beingadopted widely by the automotive industry as the state-of-the-artdefinition for vehicle functional safety at the system, hardware andsoftware levels. Building an automotive system compliant with ISO26262 is a significant task. To help mitigate risk of non-complianceand reduce development and certification costs, BlackBerryQNX provides a reliable RTOS foundation that is pre-certified tothe highest level of ISO 26262 – ASIL D. BlackBerry QNX cangreatly relieve the certification burden for automotive manufacturersand give them the peace of mind they need when buildingsystems with safety-critical requirements.Engineered for distributed processingThe autonomous car control system is a distributed processingenvironment that manages the flow of data from sensors todecision making, and finally to actuation of the physical components,such as steering, throttle and braking. Given the fast-pacedinnovation in both hardware and software, it is important to createmodular software components that can be deployed independentlyfrom the underlying hardware implementation.The QNX Platform for ADAS supports this modular framework fordistributed processing using network plug-ins for modern publishand subscribe communications technologies including DataDistribution Service (DDS), SOME/IP and Robot OS (ROS). Inaddition, time synchronization protocols such as IEEE 802.1ASand IEEE 1588 are available to provide network synchronizedtimestamps and support time synchronous control algorithms.Optimized for automotive siliconAutomotive system-on-chips (SoCs) are continually getting fasterand powerful to handle the computation load for image processing,machine learning and digital control algorithms. The QNXPlatform for ADAS software is built to be compatible with specializedprocessing cores available on a variety of ADAS processors.

Features• Reference implementations for four camera surround view, single camera ADAS, sensor hub with multi-camera input• Low latency sensor data acquisition: support for camera, radar, LiDAR, IMU, GPS sensors• Publish and subscribe sensor data access• Data visualization• Network plugins to provide sensor data over automotive networks• Sensor data capture with time stamped data samples. Sensor data playback maintains timing fidelity• Configurable timestamp sources such as IEEE 1588 PTP or IEEE 802.1AS• Robot OS (ROS) integration for testing and prototyping. Export data using ROS to compatible tools such as Matlabs• Integrated open source libraries including OpenCV, SOME/IP, Ceres and others