By guest contributor Paul Sakamoto, Chief Operations Officer, Savari, Inc.
It’s clear that Internet-connected, autonomous vehicles are no longer a pipe dream. In fact, we are already testing an assortment of connected vehicle technologies in Wind River’s “rolling lab” — a fully tricked out Honda minivan that talks to roadside infrastructure and other connected vehicles.
Along with various other technology vendors, Savari is collaborating with Wind River on vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) technology. Through the collaboration, the companies will work toward tightly integrating Savari’s dedicated short range communication (DSRC) software stack with Wind River’s portfolio of automotive software.
But this demonstration of innovative auto tech isn’t just for show — it could actually help to create a path to production for connected — and eventually autonomous — vehicles.
Putting Connected Vehicles to the Test
According to Statista, the projected size of the global market for connected technologies will be around 14.8 billion by 2030. And it’s predicted that about 98 percent of cars will be connected to the Internet by 2020. That’s just two short years from now.
Connected cars will enhance drivers’ experience with driver-assistance apps, provide enhanced navigation and entertainment services, and reduce traffic-related collisions, among other benefits.
But, as you can imagine, testing connected and autonomous vehicles presents a multitude of challenges to automakers and technology vendors. Safety, of course, is the main concern. Connected vehicles use a collection of hardware and software from different vendors — and their interoperability is essential to the safe operation of the vehicle. Networking between sensors within the vehicles, with other connected vehicles and with roadside infrastructure is also extremely complex, and can be affected by variables in the environment, such as road conditions or weather.
And the road to adequate testing is a long one. By 2016, Google’s autonomous vehicles had been road tested over 1.7 million miles and were involved in a handful of collisions, of which human drivers were at fault the vast majority of the time. But in 2016, software in one vehicle ultimately caused a crash.
Even one crash is unacceptable, particularly if autonomous cars are to be mass produced for consumer use.
Centralized or Distributed?
Malfunctioning or outdated software, or the inability of various sensors and software components to communicate within and outside of the vehicle, are key areas of focus for testing connected vehicles — and the solution may lay in the technology architecture.
Modern cars have more than 120 microprocessors, each controlling a different feature or function. For example, if power locks have a dedicated electronic control unit (ECU) that controls that function. Another ECU controls anti-lock brakes, and others control the automatic lights, interior lights, windshield wipers and so on. Although this is a vast improvement in cost and weight over hundreds of pounds of copper wire running throughout the vehicle, it still leaves a lot of room for error.
Tesla pioneered the early adoption of over-the-air updates — a necessity for enabling safe, cost-effective autonomous driving. Their cars include a small central supercomputer which controls the powertrain, steering ratios, displays and other functions. This centralized approach is now being challenged by a more distributed architecture, in which groups of functions within the vehicles are controlled together — so-called “zone computing.”
Today there’s an opportunity to re-architect in-vehicle computing, because there are more demands on the computing system than ever before. Connected vehicles require high-bandwidth communication inside the vehicle. Zone computing, runs off a data fusion hub/router, provides the benefits of over-the-air updates with redundancy, to increase safety and reduce the cost maintenance.
The rolling lab is testing new hybrid architecture, which should help all the project collaborators involved identify and troubleshoot any interoperability issues that may impede the success of the experiment.
Interoperability Enables Path to Production
The Wind River rolling lab demonstration vehicle is testing emerging technologies for connected vehicles to determine how vehicles can communicate with other vehicles and infrastructure. The minivan includes technologies such as vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, connected vehicle cockpit software, smart sensing and mapping. The experiment will help researchers and automakers visualize use cases for connected vehicles and a path to production.
For Savari, this is the first time integrating its technology into a zone-based control scheme, as it typically only ports its software to individual ECUs in distributed architectures. Savari’s application suite securely communicates between vehicles and between vehicles and the signal light infrastructure.
Unlike other sensor technologies, such as Radar, Lidar and cameras, Savari’s V2X software doesn’t require line of sight to the road, intersections or any object in order to communicate with it. This results in increased driving accuracy, especially in hazardous conditions. Vehicles can communicate with other vehicles, infrastructure and even pedestrians up to a mile away.
What types of things are communicated? Savari’s technology keeps drivers safe with collision and blind spot warnings, lane change and passing advisories, and curve speed and work zone alerts, to name a few. The software can also enable cars to identify pedestrians crossing streets in unpredictable situations, and send alerts either via smartphone or in-vehicle infotainment systems.
Because Savari’s software stack is radio-agnostic, it can fit any radio module. The software runs in the cloud so it’s not tightly coupled with a specific radio module. This will enable the V2X stack to adapt easily to future technologies, such as LTE V2X.
Wind River Keeps It Rolling
Wind River’s platform is a flexible autonomous driving platform for sensor fusion, path planning and controls/feedback. ECUs can be consolidated for secure real-time behavior. Delta technology reduces download and update times, and a plug-and-play sensor and algorithm framework speeds software/hardware integration. The platform provides advanced connectivity for OTA and SWLC management, vehicle monitoring, data aggregation and analytics. Other features include HD maps, occupant facial, emotion and gesture recognition technology, and an OS-agnostic agent for the download, verification and installation of updates to the vehicle’s ECUs.
Connecting the Dots
Savari’s solutions are also part of a massive pilot in Tampa, FL, in which 1,200 vehicles will be equipped with Savari’s on-board units and software, and data will be collected and analyzed to determine whether traffic workflows are safer and more efficient within a connected infrastructure. Many other small pilots have already illustrated the potential benefits.
And as engineers and researchers involved in advancing the viability and adoption of connected vehicles work out the kinks, we’ll move closer to a future in which vehicles operate smoothly and seamlessly on smart roadways optimized for traffic flow and safety.