Over 72.5 million connected car units are estimated to be sold by 2023, enabling nearly 70% of all passenger vehicles to actively exchange data with external sources. The amount of data resulting from these smart vehicles will be overwhelming for traditional data processing solutions to gather and analyze, as well as the associated latency of processing this data– leading to potential life-or-death scenarios, according to Ramya Ravichandar, from Foghorn.
We speak with Ravichandar, about how connected car manufacturers are implementing edge AI solutions for real-time video recognition, multi-factor authentication, and other innovative capabilities to decrease network latency and optimize data gathering, analyzing and security.
Digital Journal: What are the current trends with autonomous and connected cars?
Ramya Ravichandar: Automotive companies are looking to improve real-time functionalities and accelerate autonomous operations of passenger vehicles. Connected vehicle technology is introducing a new dimension of transportation by extending vehicle operations and controls beyond the driver to include internal networks and systems.
One of the main concerns connected cars and autonomous vehicle (AV) features aim to address is driver and road safety. Technology such as facial recognition, AI and computer vision are being implemented into smart vehicles in an effort to improve the safety and security of the driving experience.
DJ: What will the car look like and what will it be able to do in 5 or 10 years time?
Ravichandar: The connected car will continue to augment the driver experience. The future of transportation with AVs remains largely unknown, even as technology developers and manufacturers are investing in research and development.
In the near term with the rise of 5G technologies there is going to be a shift in the way the vehicles will be able to communicate with each other, and the infrastructure. An AV will become another compute platform in the network that enables a whole set of new applications. The car as we know it will become an intermediary extension of your destination; for example if going to work, the car is just a mobile office. We are already seeing trends of corporate buses being equipped for maximum worker productivity.
In about a decade the car industry will be more services driven, and there is going to be dwindling asset ownership. People will subscribe to journeys and experiences, rather than focus on physical ownership. With this changing model, the ecosystem that we know today will also change. Right from gas pump stations to mechanic shops, there are going to be new modes of engagement with the AV passenger. Large parking spaces will now be available to be repurposed to be more beneficial to the community. Cars to AVs will have a ripple effect through the system
Much of its future will depend on creating the intelligence required to build and operate AV systems. And while safety systems and high definition mapping have received a lot of attention, there are many other intelligence-intensive technologies that also need to be developed before large scale AV acceptance is possible.
As an example, nearly all AVs are expected to be electric cars and these will require substantially more in-vehicle intelligence and system life cycle management. These are needed to maximize the efficiency and lifespan of battery and charging systems as well as other systems that support braking, motor performance, safety, passenger environment and predictive maintenance. The level of compute required to manage these new systems necessitate reliable high-speed computing, which would require a significant reduction of the volume of traffic sent to the cloud, in order to lower latency.
DJ: How advanced is AI becoming, in relation to vehicles?
AI is being implemented for numerous functions of the driving experience. AI is used in vehicle functions anywhere from GPS to IoT sensors to battery monitoring, to vehicle diagnostics and even within the manufacturing process of the actual vehicle. AI capabilities have vastly advanced the capabilities of an average car and will continue to advance driver-friendly functionalities, combined with machine learning and edge computing. AI has evolved into the new norm, as nearly every car produced today has some form of AI included– it’s become the standard for traffic accident prevention and vehicle diagnostics.
DJ: What are the main challenges impacting AI and connected vehicles?
Ravichandar: Cars generate significantly more data today than ever before, and it’s become a great challenge to gather, merge, process, and deploy all that sensor data efficiently. With 72.5 million connected car units estimated to be sold by 2023, nearly 70% of all passenger vehicles will be actively exchanging data with external sources. This amount of data will be overwhelming for traditional data processing solutions, like the cloud and on-prem, to gather and analyze. This behemoth of data poses threat to clouding clear lines of vehicle-to-vehicle (V2V) communication, as well as the speed of IoT network connectivity — a potential life-or-death scenario when it comes to connected vehicles.
The future of transportation with autonomous vehicles (AV) depends on creating the required intelligence and processing to build and operate sophisticated, autonomous systems. For example, many AVs are expected to be electric cars, which will require substantially more in-vehicle intelligence to maximize the efficiency and lifespan of battery and charging systems. These features will also enable efficiency within systems supporting braking, motor performance, safety, passenger environment, and predictive maintenance.
DJ: How can the issue of data-overload be addressed?
Ravichandar:The mass amounts of data from the rise in smart vehicles are not expected to subside. If anything it will increase and continue to explode.With 5G a lot of this communication is much faster, with extremely low latency response times. However, edge-enabled AI solutions are being leveraged to more quickly and more efficiently process this data, allowing for the behemoth of data to become more manageable. Edge + 5G + AI are the key intersecting technologies that can meaningfully infer, manage and transform the massive amounts of data generated. And this is data not just from the vehicle but from the other nearby vehicles, infrastructure, and assets that are all now communicating.
This also increases the quality of insights produced from vehicle data, as there’s a larger pool of data to pull from.
DJ: How is edge computing a viable solution?
Ravichandar:Edge computing provides access to local, real-time data processing to allow for more accurate real-time insights and reduction of network latency. Edge computing addresses projected connected car challenges such as increasing road safety, increasing traffic efficiency, leveraging additional data to make connected cars “smarter” and maintaining consistent connectivity in high traffic areas. Connected car manufacturers are implementing edge computing-enabled AI solutions for real-time video recognition, multi-factor authentication, and other innovative capabilities to decrease network latency and optimize data gathering, analyzing and security.
Edge computing solutions can also help weave out the “dirty data” within the large pools of vehicle data. As a result of this, in combination with a closed-loop edge to cloud ML model, the actionable insights produced from vehicle data will become more refined and their effectiveness and accuracy will continue to improve.
While fully autonomous vehicle controls are years away, there are many existing edge computing applications now available to enhance the efficiency, reliability, and safety of commercial and public transportation. These include vehicle control and safety systems, such as cameras, driver assistance, and collision avoidance functions, that are being improved and added to new vehicles every year.
DJ: Why is edge computing superior to cloud computing in this context?
Ravichandar:Edge computing technology brings the data processing center directly within the vehicle itself, as opposed to relying on remote data centers for critical command and control decisions, and constant connectivity to these centers. Processing data at the source allows for more accurate, real-time insights and reduces network latency, which is critical when it comes to safety features within vehicles. With cloud computing, there is latency in the round trip for vehicle data to be sent to the cloud, processed, analyzed and then sent back with actionable insights. This could result in a life-threatening situation. With edge computing, automotive manufacturers can eliminate safety concerns and fast-track the road to autonomous driving by deploying edge-enabled systems within their vehicles. It must be noted that edge computing is complementary to cloud, and for emerging trends like federated learning they are in lock step.
DJ: Are there any cybersecurity concerns with this technology?
Ravichandar:Edge computing platforms can be built to operate in heterogeneous environments, simplifying the integration of a wide range of disparate systems from multiple vendors within a single vehicle. This minimizes risk of exposing data by eliminating the need to transfer data from the source to a data processing center. Edge AI not only allows data to be gathered, processed and analyzed, but also allows for actionable insights from the data to be produced at the edge — all without having to connect to the cloud.
