preloader

Search Here

INDUSTRIAL AGRICULTURE & AUTOSAR

Project Scope

Leveraging the Automotive Open System Architecture (AUTOSAR) standards, the AVISTA team helps our customer build and test AUTOSAR-compliant application components that are re-usable across families of industrial agriculture vehicles across their enterprise.

Utilizing the PREEvision modeling tool, requirements and software architectures for reusable modules can be specified at a level of precision that makes automatic code generation for the interfaces possible. The AUTOSAR standard defines the Virtual Function Bus (VFB), which represents precise interface and runnable (task) semantics between multiple application components that can be implemented on the same ECU or distributed over several networked ECUs.

Several communication paradigms are supported, including Sender/Receiver (data flow), Queued Sender/Receiver (asynchronous events/messages), and Client/Server in both synchronous (blocking) and asynchronous (unblocked) modes. Coupled with the standardization of the Run Time Environment (RTE) and Basic Software (BSW) implementations for OS interfaces, AUTOSAR services (e.g. diagnostics, NVRAM, configuration and calibration, the communication layer, etc.), IO hardware abstraction, and complex device drivers, it is possible to build real time control systems out of generated code using no hand-written code modifications.

Solution

As our customer began adopting the AUTOSAR paradigm across their enterprise, AVISTA was contracted to join this R&D effort to determine the efficacy of the chosen methodology and tool chain and uncover issues. Using a feature that had variant implementations on Ag Sprayers across the globe, a prototype of this feature was built that could be reused on every variant. Using comprehensive system requirements, several architectures and designs were built and evaluated. The chosen design was specified and wrapped within a simple application controller, and an extensive tool chain was exercised from an architecture specification using PREEvision down to bench testing using the TriCore tasking debugger and a CAN network analyzer.

Lessons learned were shared with the customer, and this effort was used as the basis for follow-on work. Indeed, that work has been seamlessly integrated within the larger architecture described below.

The customer then began a wholesale migration of existing Ag Sprayer Control functionality using AUTOSAR. The software architecture began from scratch with the intention to support all variants and even consider reusability for other non-sprayer types of vehicles. The AVISTA team proposed a layered architecture approach separating user interface dependencies using the proven Model-View-Controller approach. Business logic for all variant features was separated from physical sensor monitoring and actuator control. Models of the plumbing and propulsion (pumps) system were designed to isolate plumbing changes from the feature logic and user interface. The architecture was designed to support increased Sprayer/Vehicle automation.

During the prototype effort, AVISTA investigated issues with hybrid combinations of AUTOSAR- and non-AUTOSAR-compliant models and code generation issues. AVISTA quickly demonstrated what modeling capabilities (and resulting generated code) could be shared and what capabilities are dependent on RTE/BSW tasking and memory models (e.g. NVRAM). Due to the incompatibility of the middleware and OS layers, the customer shifted their hybrid strategy to other solutions.
Working with the customer’s team (relatively inexperienced with MBD), AVISTA helped develop the functional components identified in the new architecture and helped train the customer in optimal modeling techniques and unit test simulations.
AVISTA pioneered innovative testing strategies and solutions to rapidly create module tests capable of exhaustive testing of modal control system functionality. Libraries of test harness functions were built that enabled test scripts to be written in tabular form, so all combinations of input conditions and resulting expectations clearly showed test intentions. These tests formed the suite of Continuous Integration tests. Coupled with Simulink’s Model Code Coverage tools, modified condition/decision test coverage was analyzed and documented.
AVISTA’s passion for testing drove the development of a large software integration MIL environment containing all AUTOSAR-compliant components. Initially, open loop simulation verifies successful feature behavior as all components are cooperating. Subsequently, a plant model of a sprayer (pumps, tanks, valves, etc.) is integrated in this MIL environment to ensure the software meets system requirements. This MIL is also used during the Systems and ECU Integration phases to corroborate functionality expected on the physical controller connected to electrical interfaces.
AVISTA pushes through technical challenges posed by tool chain issues. Despite the AUTOSAR partnership being over 15 years old, tool chains supporting the AUTOSAR methodology are non-trivial in their configuration and sometimes inconsistent in their support for various AUTOSAR features. For example, full support for AUTOSAR sender/receiver interface compatibility rules permitting receivers to be connected to parameter and non-volatile data interfaces is not fully supported. This limits flexibility during Systems Integration. Despite tool limitations, AVISTA uses a best-effort approach to keep moving forward.