By Vincent Roche, President & CEO, Analog Devices
Every year we harvest and create ever-increasing quantities of data. By some analyses, we are generating 2.5 quintillion bytes of data every day and 90% of the data we have today was created in the past two years. This data deluge shows no sign of slowing as the Internet of Things and applications operating on the edges of networks become more prevalent. To transform this data into information at the point of creation and need, and to do so more efficiently, processing is increasingly being decentralized and we’re seeing an upsurge in the growth of edge-node processing.
In 2019, we’ll continue to see advances in “smart sensors” that only wake when something interesting is happening and sift the input to only transmit relevant information to the cloud. We’ll also see greater focus on efficient power management in edge node devices to support the increased processing demands and resultant thermal dynamics challenges. We’ll see greater focus on security as the growing value of the information generated makes loss or corruption more costly. Finally, sensor fusion, or the combination of differing sensing modalities to create a more accurate perception of reality, will become increasingly pervasive thanks to a combination of edge-node processing, the fog, and the cloud.
Artificial Intelligence Begins to Shed the Artifice
It is difficult to imagine a technology more hyped at this point than Artificial Intelligence/Machine Learning (AI/ML) but the reality is that below the breathless excitement, real progress is being made as virtually every industry is working to determine the opportunity and potential impact it represents. In automotive for example, rapid advancement in environmental sensing accuracy and the delivery of higher quality, more relevant data and information to AI engines is fueling progress in both trained and inference-based systems.
Automotive AI processing predominantly takes place at the edge rather than in the cloud due to the need for low-latency, real-time decision making but the architectural battle between edge and cloud computing in AI will be waged on an application-by-application basis as other emerging markets such as gaming and industrial automation have their own, different needs. Thorny issues remain for all markets, especially in the areas of unsupervised training and validation of how AI systems will really work once they are trained. While we will make advancements in these areas in 2019, non-technical issues like ethics, liability, and governance as well as a limited pool of AI/ML talent remain hurdles to progress.
Shifting into High Gear
A triple wave of disruption is hitting the automotive industry – vehicle electrification, autonomous vehicles, and transport-as-a-service. Vehicle electrification will continue to grow at double digits and migrate from the niche to the mainstream in 2019 as internal combustion engines face increasing regulatory pressure (up to and including banning certain technologies) and electric and power management technologies improve and become less expensive. New chemistries are enabling much greater energy density in batteries, which is spawning innovation in related areas of battery formation and lifecycle management. Advancements in electronics and architectures will be necessary to enable the sensing, measurement, interpretation, and secure communication of battery state information that is needed to maximize the results, lifetimes, and safety of the next generation of batteries.
On the autonomous vehicle side, we expect to see additional advances in level 3+ highway autopilot solutions, extensive tests of fleet services/robotaxis, and the fusion of vision, RADAR, and LiDAR modalities to enable greater situational awareness in autonomous vehicles. Robotaxis from companies like Baidu and Waymo will bring the next wave of innovation to the transportation-as-a-service market and continue to drive the consumer preference shift from car ownership to car access.
Significant progress is being made in the area of autonomy and those advances aren’t limited to the automotive market. Robots, and increasingly, cobots (human-robot pairings) are operating independently of programming more and more as they are equipped with sensors and local, fog, and cloud-based processing power that enables them to rapidly learn and adapt to their environments and circumstances. In 2019, we will see additional advances in the security and safety of autonomous electronics in areas in which humans and robots interact. For example, autonomous robots and cobots able to map their surroundings in 3D through multiple modalities (e.g., vision, LiDAR, time-of-flight, etc.) in real time will become increasingly important as they work in close proximity to humans and need to adjust their operation when human safety is threatened by that operation.
Healthcare Becomes Prescient
Spiraling healthcare costs and growing disease prevalence spurred by the effects of modernity are creating a new level of urgency to solve the larger healthcare challenge. The need for economically viable wellness and healthcare solutions will drive strong growth in the adoption of sensing, computing, and mobile technologies. Deeper diagnostics, miniaturization, and connectivity will be the guideposts on the path to healthcare that is increasingly prescient.
As a consequence of this disruption, we are witnessing the birth of a new healthcare era in which clinical-grade healthcare will be available in non-clinical locations and applications. Biological, gas, and chemical sensors will increasingly provide clinical-grade measurements in form factors and price points that make them more accessible to applications outside of traditional healthcare facilities. When combined with advances in mobility/connectivity and cloud computing, these advances are making it possible to manage patient diseases in ways that prevent acute events that require expensive hospitalization, not to mention life disruption.
In 2019, we will see more clinical-grade consumer devices being used in healthcare regimens as insurance companies, hospitals, and government programs like Medicare start to accept this as a valid way to monitor patients and improve healthcare outcomes. This will drive additional advancements in sensor accuracy and data analysis as at-home devices begin to transition from a recreational novelty to a truly critical element of healthcare.
Can You Hear Me, Now?
5G, which has been on the horizon for many years, is finally undergoing substantial testing and will begin to be commercially deployed in 2019 if not by the end of 2018. The initial phase of Massive MIMO (Multiple Input, Multiple Output) will provide significant improvements over 4G and eventually, millimeter wave 5G solutions will provide multi-gigabit per second wireless connectivity. As these initial 5G instantiations come online, the impact will be substantial.
Just as previous generations of wireless connectivity have created new markets and business models (e.g., mobile e-commerce, video streaming, etc.), the bandwidth explosion, low-latency speed and responsiveness, and highly configurable network solutions brought by 5G will dramatically remake markets and open new ones. Unlike in previous generations, however, the advances in connectivity enabled by 5G will extend beyond the internet to drive revolutionary changes in automotive, healthcare, and industrial automation markets.
Ultimately, one of the only predictions we can make with certainty is that 2019 will evolve in ways we can’t predict. Some technologies that seem on the verge of achieving their promise at the start of 2019 will still be on the verge of achieving their promise at the end of 2019. Other technologies will silently reach a tipping point and become part of our everyday experience with hardly a ripple.