Advanced driver assist applications require adoption of demanding safety standards

4 min read

Safety is an increasingly important topic for system developers and there is an ever widening range of safety standards that developers need to be aware of. The increasing use of electronics in cars has led to new safety standards and new demands on developers of both hardware and software.

Devices need to be secure as potential attacks could result in undermining safety functions. As a result both safety and security standards are taking centre stage and need to be taken into account when it comes to next generation designs; recalling a range of vehicles can be very costly and damaging to the car maker's brand.

The automotive industry is a growing market for microprocessors as more vehicles are designed incorporating embedded computer systems to address safety and efficiency demands, whether in response to legislators or the demands of consumers.

Research firm IC Insights expects semiconductor consumption in the automotive industry to grow at an annual rate of 11% until 2018, driven by the demand for cameras, eCall, chassis control, instrument cluster processing and other advanced driver assistance systems (ADAS).

In January, ARM unveiled a comprehensive safety document set for the ARM Cortex-R5 processor in a move designed to drive its adoption in safety-critical applications.

According to Richard York, VP Embedded Marketing, ARM: "It is generally accepted that cars are fast becoming a supercomputer on wheels, whether that involves driver assistance capabilities or the move towards semi-autonomous driving. All of these applications demand greater processing capabilities in order to make decisions that could affect the vehicle's stability."

The delivery of the safety document set for the Cortex-R5 processor was seen as a vital step toward the cost-effective deployment of more technically advanced systems across multiple sectors.

"This is the result of more than three years of investment in internal processes, training, verification and documentation flows, requirements tracking and safety management," explains York. "The automotive market is demanding multi-sourced, powerful and standardized MCUs that are supported by well established tool chains. Safety is the critical success factor."

"Functional safety is increasingly important for these rapidly-expanding markets," suggests Noel Hurley, general manager, CPU group, ARM. "The Cortex-R5 processor has been developed with an extensive set of fault detection and control features and the addition of generic safety documentation that will mean that developers can now use it across the broadest range of safety applications."

For silicon vendors looking to address the automotive space functional safety is critical, but has often been seen as a cost.

"Functional safety techniques are being applied at increasingly lower levels of design abstraction," suggests Norbert Asche, general manager, safety microcontrollers, Texas Instruments. "Our Hercules TMS570LC4x and RM57x microcontrollers (MCUs), which are based on the ARM Cortex-R5 core, as well as our SafeTI design packages are intended to help designers meet standard functional safety standards such as ISO 26262, IEC 61508 and ISO 13849 while at the same time managing both systematic and random failures."

According to York there is a strong focus on advanced driver assistance in end applications but whatever the application, whether collision avoidance, driver cockpit IVI or lane detection or steer, there's a powerful processor taking data from sensors and a higher level processor making the decisions.

Earlier this year Freescale looked to address the market with the launch of the AS32V processor family for advanced driver assistance systems.

The S32V family has been designed to provide ADAS computational needs with a mix of CPUs, GPU, and image cognitive processors and, importantly, it does this while maintaining a power envelope of just 5W. Low power is critical to automotive systems as they should be passively cooled as it is difficult for a fan-cooled system to meet automotive system reliability standards.

The S32V234 addresses the optical processing requirements of four cameras with dual CogniVue APEX2 processors and the GPU is used to build a surround view of the environment around the car: building a real-time 3D spatial model of the environment around the car and providing object and threat calculations.

"ADAS is emerging as a key technology in the automotive space, employing video cameras or radar systems to provide the driver with additional information and, in some cases, to automatically take some degree of control of the vehicle powertrain or chassis. These systems require a large amount of computation capability and ADAS needs software development solutions that will help accelerate system software integration and test," explains York.

"In the modern vehicle, all of these systems are interconnected using a dedicated and robust communications network designed specifically for automotive such as CAN, LIN, FlexRay or TTP."

The car area networks transfer data and control signals between hundreds of ECUs that are located around the car and the various sensors and actuators they control. Each ECU will have inside one or more microcontrollers, many of which will contain a processor of some other kind and each ECU will contain a program memory with many kilobytes of software programmed into it.

Overall, the cost and complexity of automobile electronics is rapidly escalating as next generation ADAS and other systems are introduced.

"Processors for the automotive space need to be able to make decisions based on information that will be complex and will need to compensate for external problems such as driving rain, fog, glare and reflection, as well as the ability to effectively monitor lanes. Talk to software developers and they'll take as much computing power as you can supply," says York.

With so much processing the number of sensors on board is soaring.

"The amount of data being pumped into control units to make decisions is growing and we expect to see a 100 fold increase over the coming ten years as assistive driving, such as collision avoidance steering and advanced camera systems give way to advanced, all-round collision avoidance and limited autonomous driving, which will then give way to autonomous driving which could involve high speed all-round collision avoidance and connected vehicle to vehicle," York suggests.

"By 2016, the majority of consumers in mature markets will consider in-vehicle web-based data access a key criterion in their automotive purchase," explains Thilo Koslowski, VP of Automotive at Gartner. "Successful connected vehicle solutions will add value to the connected driver's digital lifestyle and enable more integrated cross-device experiences."

As if to prove that point, Nokia has recently developed a new connected car platform called Here Auto. An embedded platform, it comprises of an application that can be embedded to a vehicle's navigation system.

Described as a complete Connected Driving offer, it is intended to help car makers and in-vehicle technology suppliers connect the car and the driver to the cloud and, by doing so, provide a real-time driving experience that will better understand current driving conditions. Here Auto is being deployed by Jaguar Land Rover.

Technology is rapidly transforming driving, with connectivity bringing the outside world to the car and the sensor revolution creating new possibilities for the automobile industry.