ALPHA-AMV automotive development platform
Connectivity
- Car2X
- IoT
- Over The Air (OTA) software upgrade
Dynamic Software
- Platform
- Adaptive AUTOSAR 18-03
Stack to scale
- Add more boards for more processing power
Truly heterogeneous
- Unified SoCs
- PCI-E connectivity
- Inter-SoC vision SDK
Powerful platform
- 3 x TDA2x
- 10 x camera
- Linux / RTOS
- Rich interface set
Introduction
At the very beggining of its engagement in the field of camera based autonomus driving, RT-RK invested heavily in getting deep knowledge about the systems, state of the art solutions, algorythms, problems and spaces to participate in improvements.
Researching possible hardware platforms, applications requirements, processing power needs, limits and constraints leaded us to an idea to build our own development and education platform, suitable to acumulate knowledge, learn and demonstrate new developed solutions.
Such idea was warmly welcomed by silicon vendors, Texas Instruments before others, and we defined the scope for the project: develop an automotive vision reference design board, on Texas Instruments chipset, that will support basic and advanced warning systems, active controll systems and semiautonomus applications.
A set of targeted aplpications covered:
- Surround view
- Driver monitoring
- Stereoscopic front view
- Wide angle camera
- Night vision
Target audience:
- Car manufacturers (OEMs)
- Automotive design houses (Tier1/2/3) for rapid prototyping
- TI advanced automotive customers
- Algorithm developers for demo purposes
The result of the development was very powerfull system, based on triple TDA2x Soc from Texas instruments, suitable for up to 10 cameras, with load balancing for video processing algorithms between SoCs via PCIe, and triple HDMI outputs as links to the head units, as well as 1Gb ethernet connectivity supporting AVB.
As a basis to demonstrate dynamic software platform including adaptive AUTOSAR 18-03, IoT vehicle connectivity, over the air software upgrade as well as SoC Vision SDK was rounded with extensieve operation memory map, in 21 memory IC on board, each running at 1.6 Gbps.
Genuine heterogenuity was accomplished by stack interconnectivity interfaces, that enabling system and functional expansion, including automotive functional safety blocks and sensor fusion applications.
Conclusion
Truly research and development projects with silicon vendors included, such as ALPHA-AMV, bring cutting egde techology to our desks and require finding solutions beyond everyday life praxis. Besides the announced and defined results of the projects, providing such compact, easy manufacturable yet powerfull development and demonstration platform before wide availability on the market, paved the way to put ourselves on a map of autonomus driving asistance knowledge providers.