System Architecture Team
App Framework and Security
Continuous Integration and Test
UI and Graphics
Reference Hardware System Architecture
Vehicle to Cloud
System Architecture Team
App Framework and Security
Continuous Integration and Test
UI and Graphics
Reference Hardware System Architecture
Vehicle to Cloud
Below is a table with the consolidated collection of project proposals. Members of the AGL Steering Committee are encouraged to rate the proposals so that we get a priority ranking. You need a login to rate since your submission will be traced and you can only rate a proposal once.
|1||05/13/14|| Improved Build System ||Improve the draft Yocto based build for Tizen IVI to support full product builds of the AGL stack. The objective of this project is to be able to build Tizen IVI stack using the Yocto tools in a fashion that is compatible with the current development workflow that is based on OBS/GBS. An early feasibility study has already been conducted, the work is in progress but there is still work to be done. This wiki page describes the current state of affair: https://wiki.tizen.org/wiki/Tizen_on_yocto. The development is done on Tizen Gerrit and all the code is updated directly there: https://review.tizen.org/gerrit/#/admin/projects/scm/bb/meta-tizen.||The current OBS+GBS based build system is cumbersome to use. Organizations who do not wish to learn and set up the required OBS infra for full builds are not able to perform full AGL builds with GBS only (see https://bugs.tizen.org/jira/browse/DEVT-165). GBS was never designed to manage full dependency chains, it is designed to perform only individual package builds, so this is not likely to be fully fixed in the context of GBS.|| Reaktor |
| Juha-Matti Liukkonen |
Dominique Le Foll
|This project enables AGL developers to build the full system from source (for x86 and arm), without the need to learn or set up OBS infrastructure. The AGL community will have the opportunity to choose their preferred build tools (Yocto or GBS/OBS) for building the AGL reference distribution.||Duplicate of 11? [RJS 2014-08-13] Consolidated this with 11 and removed 11.|
|2||05/13/14|| AGL parity on ARM |
ARM Platform Support
| Improve the ARM support in AGL for a widely available ARM platform (probably Renesas) to enable x86 and ARM feature and maturity parity. |
Add ARM architecture (Renesas R-Car M2) support to AGL. As a first step, we work with Tizen project to add ARM support for newly announced “Tizen common” code base. It would be shared common ground also for AGL project. Once AGL low-level components adequately defined, we will fill the gap if any.
| Due to unit cost, OEMs aiming to productize AGL on volume market models need to have ARM based IVI platforms available as option to x86. Currently Tizen IVI development takes place almost completely on x86, although the underlying Tizen platform supports also ARM. Tier-1s have clearly stated that they do not wish to use multiple platforms (for high-end / volume market), which means that as long as AGL is limited to high-end x86, its adaptation for commercial use is limited. |
Current AGL assets rely on Tizen IVI (kernel, driver and framework), and some part of Tizen IVI is designed for Intel CPU/Graphics architecture. To add ARM support, we need to migrate Tizen-common code base as it already includes architecture agnostic porting layer. In AGL, we will validate all application on top of R-Car environment.
| Reaktor |
| Juha-Matti Liukkonen |
| This project improves the commercial feasibility of using AGL in real vehicles. |
[RJS 2014-08-14 Merged 34 into this and removed 34.
|3||05/13/14|| Full prototype |
Full Demonstrator Development
| Implement a realistic IVI system prototype using AGL and all of its key features (HTML5, SDL, murphy, amb, smack, etc.), preferably providing feature parity on both x86 and arm based systems. |
AGL full demonstrator can emphasize the value of adopting AGL as a foundation software environment to support modern IVI system. It should contain attractive feature for IVI system like fastboot and/or quick rear view camera startup feature. We will provide an open platform (low-cost reference board + fully open software) so that anyone can play with AGL code and study it to develop a real product. Renesas can work with AGL member companies to bring up R-Car M2 based AGL demonstrator environment.
| Many open questions from Tier-1s relate to integration of native navigation engine to HTML5 user interface, integration of voice recognition, integration of mobile connectivity, performance and memory use of HTML5, and so on. These issues cannot be measured or known with certainty until realistic reference implementations exist. JLR has been doing great work in advancing the AGL stack to provide measurable implementations of advanced use cases. |
AGL is designed for IVI and other automotive systems that contain some specific demands. Therefore, it is not easy to try on regular PC environment to fully validate supported features. The objective of creating AGL demonstrator is to provide out-of-box experience to the automotive industry people and penetrate AGL asset to wider automotive use case.
| Reaktor |
| Juha-Matti Liukkonen |
|This project provides OEMs and Tier-1s, and all developers also, a concrete and measurable reference implementation of all of the key integrations required for building commercial products.|| Not sure what concrete or measurable mean? It could be overlap with Tizen IVI? Could also be overlap with GENIVI baseline - meta-ivi?
[RJS 2014-08-14] Merged 35 into this and removed 35.
|4||05/13/14||LirrorMink||Reference implementation of LirrorMink.||This would require some homework first, though — the CCC is a rather lawyer-happy organization. They may not be OK with an open source reference implementation of something-that-isn’t-called-MirrorLink but that would be compatible for stack integration / verification purposes.||Reaktor||Juha-Matti Liukkonen|
|5||05/13/14|| Automotive Hypervisor based on Xen project |
Develop XenGT into an automotive-grade hypervisor
| Developing and maintaining automotive grade hypervisor based on Xen project. Ability to run different type of guest OSes: Android, Tizen, QNX, Autosar, Linux. Support A15/A50 SoCs - TI J6, Renesas Car-M2, Qualcomm, Freescale i.MX 8 |
The ability to consolidate workloads in automotive presents a very attractive value-proposition, but it raises many questions and concerns at the same time. Virtualization is seen as a key technology enabler that can address many of these considerations. What XenGT specifically brings to the table is a type-I hypervisor that offers a very efficient mechanism for sharing the GPU between virtual machine. This is a key requirement to be able to run the instrument cluster and in-vehicle infotainment systems as two distinct virtual machine for example.
|Type 1 Hypervisor for Cortex A15/A50 SoCs based IVI systems. There are strong demand from car vendors to introduce “integrated cockpit” concept, i.e. ability to consolidate IVI, ADAS and instrumental cluster on the single SoC. The virtualisation resolve the issues related to security and stability appeared by using different OSs and requirements for IVI, ADAS and instrumental cluster.|| GlobalLogic |
| Alex Agizim |
Geoffrey Van Cutsem
| This project is provide flexibility for AGL to build “integrated cockpit” solutions and feasibility of using AGL in real vehicle. |
Deliver an open-source type-I hypervisor that will allow consolidating GPU-intensive workloads such as Instrument Clusters, In-Vehicle Infotainment and potentially (soft)-ADAS functions.
| Curious to know more. What is the second OS in this context? Would it cater for share graphics across OS, and if so, what graphics display technology - Wayland and/or X11 and/or FB?
[RJS 2014-08-14] Consolidated this with 12 and removed 12.
|6||05/16/14||Robustly Availability and Serviceability||We believe the need for a mechanism in IVI system such as the following: -When the application is abnormally terminated, the system tries to recover is allowed to restart application. -When the middleware function is abnormally terminated, the system tries to recover is allowed to restart all applications that depend on the function middleware. -When a critical process is aborted, the system has to restart the system immediately to recover own. -When the process repeats abnormal termination, the system must be restored by deleting its temporary files. -When abnormal system restart occur in running system (including system update), must be able to recover at the next boot up. We propose this system to AGL as a way to robust IVI system.||Aisin AW||Naoto Yamaguchi||This project provides the robustness for AGL platform.||Appears to be focused on similar requirements that the GENIVI Node state manager and Node startup controller focus on? Recommend talking to the GENIVI project to see if there is synergy.|
|7||05/16/14||GPS time and Monotonic time support||With Linux, time adjusting is generally done by using date command or NTP but since in-car device can obtain the accurate time from GPS, GPS time is often used as Abs Time. Because of its advantage where this GPS demand can be done anywhere in the world, it would continue in future. Therefore, we are going to need a structure for adjusting the Linux system time. As a weak point of GPS, when cold start, it takes a long time to obtain the accurate time.||Aisin AW||Naoto Yamaguchi||This project helps software development of the time for in-vehicle system.|
|8||05/16/14||System Lifecycle||It watches a state of the hardware of the in-vehicle system and pulls the trigger such as application lifecycle depending on the state. So it provide specific IVI system lifecycle, such as power-On/Off from other ECU and others.||Aisin AW||Naoto Yamaguchi||This project provides the specific IVI system lifecycle for AGL platform.||Appears to be focused on similar requirements that the GENIVI Node state manager and Node startup controller focus on? Recommend talking to the GENIVI project to see if there is synergy.|
|9||05/19/14||Proposal: group based cpu resource control(bind specified processes to a cpu subset)||The AGL should provide a mechanism for assigning a set of CPUs to a set of tasks (cpuset in kernel) According to policy based decision, Policy Manager must assign proper CPU subset to target process or process group by using “Resource Control” of kernel layer. (typically cgroups/cpuset will be used)||The hardware may have several CPUs, and the system may runs several tasks/APPs. The system will benefit from careful processor placement to reduce schedule and contention(cache …etc).||Fujitsu TEN||Lai Jiangshan - SRCU maintainer. Kernel expert in RCU, virtualization, scheduler, cgroup etc..||Reduce schedule and contention.||Appears to be focused on similar requirements that the GENIVI Node state manager and Node startup controller focus on? Node System Resouce Manager seems to overlap here. Recommend talking to the GENIVI project to see if there is synergy.|
|10||05/19/14||Realtime Support for AGL||Some processes should have higher priority than others, and can preempt others immediately. IOW, the AGL should have realtime support.||Linux kernel is a Time-Sharing OS, not a realtime one.||Fujitsu TEN||Lai Jiangshan - SRCU maintainer. Kernel expert in RCU, virtualization, scheduler, cgroup etc..||Enhance latency performance for latency sensitive process.||At a macro level, this looks like focusing on kernel features such as the rt scheduler and perhaps cgroups. As it is managed, it one could predict this moves into resource management. Again, I feel overlap here with the GENIVI Lifecycle projects - worth investigating.|
|13||05/19/14||Develop and integrate Ethernet AVB in Tizen IVI||Tizen IVI already provides the foundation (adopted from the Open-AVB project) to support Ethernet AVB for IVI systems. This project will focus on what is missing to have a complete Ethernet AVB stack and look at the integration required within the system (e.g. multimedia framework, policy engine, security framework, application framework, etc.)||Intel Corporation||Geoffrey Van Cutsem||Ethernet AVB offers a cost-effective transport solution for media sources (audio, video) within the car and paves the way to a broader adoption of Ethernet as a consolidated networking technology.||GENIVI has a project running specifically on AVB which looks quite advanced in terms of both thougth process and network with Avnu.|
|14||05/19/14||Enable (soft) ADAS through OpenCV and the HW-abstracted OpenCL backend||ADAS functions are developing really fast and becoming more and more pervasive. The development process is shifting as a result and many are looking at common frameworks that they can leverage to enhance and develop their portfolio of ADAS functions. Many of these require a lot of image/video processing (e.g.: Driver Drowsiness Detection, Pedestrian , Lane Departure Warning, Road Sign Recognition, Pedestrian Detection System, etc) and would benefit from having access to the GPU. OpenCV has the ability to use OpenCL which is a Khronos standard and is implemented by a variety of devices (CPUs, GPUs, FPGAs, ARM), abstracting the exact hardware details, while enabling vendors to provide native implementation for maximal acceleration on their hardware. This project will focus on automotive use-cases and look at enhancing the existing backend to provide a more optimized framework.||Intel Corporation||Geoffrey Van Cutsem||Propose and promote a framework that can boost the development of ADAS functions.|
|15||05/19/14||Framework to add private APIs to Automotive Message Broker||The Automotive Message Broker offers a framework that exposes vehicle data in a well-defined API and serves as the foundation for the development of the W3C Automotive Web API. Each and every car manufacturer also has a number of signals that are ‘private’ or specific to their lines of vehicle. This project proposes to define and implement a mechanism that would allow to extend AMB’s current API set by adding private APIs without modifying the code source of AMB.||Intel Corporation||Geoffrey Van Cutsem Kevron Rees||Enhance AMB and make it easy for Tier 1s/Car manufacturers to adopt it and adjust it to meet their specific needs.|
|16||05/19/14||Persistency and reliability of flash-based memory devices (eMMC, SSD)||Flash-based memory devices are becoming more and more pervasive. This technology introduces many opportunities but also raises some questions notably in the areas of persistency and reliability. This project proposes to study these aspects in the context of automotive and identify gaps and/or propose solutions to address any shortcomings.||Intel Corporation||Geoffrey Van Cutsem||Set of recommendations and tools to facilitate the adoption of flash-based memory devices.||GENIVI has a long running project in this area - Persistency - a code base exists too. Managed under the GENIVI SI group.|
|17||05/19/14||Home Screen with Multi-Display||Soft PF for In-vehicle cockpit system||Toyota||Takeshi Hoshina|
|18||06/04/14||IVI Layer Management||Define a common API for IVI Layer Management Service.||In IVI each applications such as navigation system, rearview camera mostly are implemented standalone therefore it needs to composite all applications to final image.||DENSO CORPORATION||Tatsushi Yamamoto||Expect to be based on GENIVI IVI Layer Manager|
|19||06/04/14||IVI Audio Management||A framework to manage audio in the IVI.||To support automotive specific usecase for audio management like handling interrupt voice guidance at an intersection.||DENSO CORPORATION||Tatsushi Yamamoto||Expect to be based on GENIVI Audio Manager|
|20||06/04/14||Diagnostic Log and Trace||Standardized logging and tracing.||To consolidate the existing variety of logging and tracing protcols on one format.||DENSO CORPORATION||Tatsushi Yamamoto||Expect to be based on GENIVI DLT|
|21||06/04/14||NavIt & Open Street Maps||Open Source Navigation Engine & Maps. Complete with HTML5 GUI.||To create an open source option for the community, to prototype on and eventually to be production ready.||JLR||Pete Popov||Possibly adopt GENIVI LBS API|
|22||06/04/14||Dead Reckoning engine||Improving the positional accuracy using vehcile and gyro signals||To be used with NavIt POC.||JLR||Pete Popov|
|23||06/04/14||Media Player||Full media player, working with GENIVI. Including media management indexing.||Based on Crosswalk and Tizen.||JLR||Paul Wiles|
|24||06/04/14||DOORS Requirement Database||Open automotive requirements database, tracing from use cases to APIs.||Full system engineering approach.||JLR||Matt Jones||Tracability of requirements against code and modules.Change control on the specifications|
|25||06/04/14||Handwriting Recognition||Project to create open source implementation for handwriting recognition.||Important input medium||JLR||Richard Rowe|
|26||06/04/14||Speech Recognition||Project to create open source implementation for speech recognition.||Important input medium, very closed today||JLR||Richard Rowe|
|27||06/04/14||Gesture Keyboards||Flexi type keyboard integration to improve keyvboard capabilities.||Important input medium, very closed today||JLR||Richard Rowe|
|28||06/04/14||Open Telematics Architecture||Build out on board and offboard architecture for Remote Vehicle Interactions, Big Data and Software Updates.||JLR||Magnus Feuer||Expert Group formed.|
|29||06/04/14||NFC||Near Field Communiction examples and automotive use cases||NearD is used today, but not fully working.||JLR||Paul Wheller|
|30||06/04/14||Vehicle Signal Simulator||Vehicle signal simulator to enable recordings to be made, and scripts to be run. Interfacing to AMB.||Enables others to use projects without car.||JLR||Paul Wheller|
|31||06/04/14||Weather||Offboard weather service integration.||JLR||Richard Rowe|
|32||06/04/14||Enables email functionality on the head unit. Basic building block.||JLR||Richard Rowe|
|33||06/10/14||meta-agl||Create and maintain a Yocto Layer for building the AGL reference stack||Output from #1 and #11 should provide us some information as to how one creates recipes and layers from Tizen that can be incorporated into Yocto. To build on this I believe we need a viable maintenance strategy for a AGL “layer” that we can combine with meta-ivi and layers and recipes from various silicon vendors.||Pelagicore||Jeremiah Foster||New maintained layer that can be integrated into existing Yocto builds and serve as the basis of a Yocto built AGL stack.|
|36||06/23/14||AGL Automated Testing & Reporting||As the value of AGL, we should perform validation and provide test result to the public. This proposal is build automated system test framework for AGL components. This test can cover both kernel and userland. We can utilize existing test program and benchmark program as well as custom test script. We can use this facility for periodic test like nightly test and track performance trends.||Linux Foundation LTSI kernel project develops automated kernel validation suites to confirm LTS to LTSI regression. And we can utilize this infrastructure to validate AGL asset. We will develop dedicated test program to validate AGL code that can run on top of the existing LTSI test environment.||Renesas||Hisao Munakata||As a value adds to AGL, we can introduce validation result with release. Anyone want to adopt AGL can refer this test result and eliminate duplicated validation process.|
|37||06/23/14||Add V4L2 Decoder Support to Ozone-Wayland||Add V4L2 decoder support to Ozone-Wayland to enable HW accelerated video decoding for platforms that supports HW accelerated video decoding via V4L2 API.||Chrome make use of V4L2 VDA and VEA for ARM based SoC such as Exynos and Tegra4 . It is cruicial to support this capability to expand the multimedia support on wider range of SoC on Wayland.  https://docs.google.com/a/chromium.org/document/d/1LUXNNv1CXkuQRj_2Qg79WUsPDLKfOUboi1IWfX2dyQE/preview#heading=h.c4hwvr7uzkfl||Renesas||Hisao Munakata||Can expand the range of platforms that support multimedia capability on Blink/Wayland environment.|
|38||06/24/14||Device APIs for HTML5 Application||Implement automotive extension device APIs which does not depend on specified HTML5 engine.||We need automotive extension device APIs for HTML5 application. Implementation of device APIs should not depend on specified HTML5 engine, because HTML5 engine is just evolving and OEMs/Tier-1s should have the opportunity to choose HTML5 engine.||Panasonic||Takehiko Yasuda||This project provides extendibility for HTML5 engine.|
|39||06/26/14||Google Auto Link Integration||Integrate Google Auto Link into the AGL reference platform.||Symphony Teleca||Dean Miles||Bringing Google Auto Link connectivity to AGL reference platform.|
|40||06/26/14||Tizen Gap Analysis||Analyze gaps in reference AGL platform (Tizen 3.0) to a typical Automotive IVI feature requirements||Symphony Teleca||Dean Miles||Give a better view to AGL platform adopters on what is missing in the AGL reference platform|
|41||06/26/14||Fast Boot support for AGL reference platform||Improve boot time for the AGL reference platform to get closer to the automotive OEM's requirements of 2 sec for camera and 1 sec for audio||Symphony Teleca||Dean Miles||Make AGL reference platform more attractive for automotive OEMs by addressing one of the biggest concerns from customer satisfaction and regulatory requirements perspective|