Search Results

The transition to 5G is underway and is realizing the 5G future is viewed by many as essential to sustain national competitiveness and as an essential infrastructure platform for supporting Smart-X, where X may be replaced with healthcare, greener energy grids, transport systems, supply chains, etcetera. But separating the hype from reality is challenging, in part, because 5G is part of a horizon vision that has the potential to significantly alter the fundamental economics that have characterized the evolution of mobile networking through its first four generations. None of today's 5G offerings nor those that will be available in the next few years will deliver the full complement of 5G promised performance improvements, and many analysts are skeptical that those performance improvements are really needed. Key implications of realizing the 5G promise are the need to transition toward more localized and granular control of network resources and increased converged/shared ownership of core resources, but how these may be managed and who will bear responsibility for the investment leaves the implications for the competitive landscape for wireless infrastructure services uncertain. While the incumbent mobile network operators (MNOs) are likely to continue to lead the drive toward 5G, their role in the wireless ecosystem may change significantly. This talk will focus on highlighting an economists' perspective on what 5G, viewed as a horizon vision, may mean for the evolution of wireless broadband networking and the disruptive potential that vision portends.

5G and future Gs contend with an increasingly dynamic and heterogenous environment, with a multitude of vendors, wireless connectivity standards, requirements, verticals and use cases. The traditional approach inherited from telephony has been one based on careful, deterministic management. We present how such randomness, far from being detrimental, can beneficial when correctly exploited, and show that, surprisingly, random approaches in many cases are actually optimal.

This talk will describe some of the challenges and opportunities in autonomy research today, with a focus on trends and lessons in self-driving research. We will discuss some of the major challenges and research opportunities in self-driving, including building and maintaining high-resolution maps, interacting with humans both inside and outside of vehicles, dealing with adverse weather, and achieving sufficiently high detection with low probabilities of false alarms in challenging settings. We will review the different approaches to automated driving, including SAE Level 2 and SAE Level 4 systems, as well as the Toyota Guardian approach, which flips the conventional mindset from having the human guard the AI (as in SAE Level 2 systems) to instead using AI to guard the human driver. We will discuss research opportunities in mapping, localization, perception, prediction, and planning and control to realize improved safety through advanced automation in the future.

Robot perception and computer vision have witnessed an unprecedented progress in the last decade. Robots and autonomous vehicles are now able to detect objects, localize them, and create large-scale maps of an unknown environment, which are crucial capabilities for navigation and manipulation. Despite these advances, both researchers and practitioners are well aware of the brittleness of current perception systems, and a large gap still separates robot and human perception. While many applications can afford occasional failures (e.g., AR/VR, domestic robotics), high-integrity autonomous systems (including self-driving vehicles) demand a new generation of algorithms. This talk discusses two efforts targeted at bridging this gap. The first focuses on robustness: I present recent advances in the design of certifiable perception algorithms that are robust to extreme amounts of outliers and afford performance guarantees. These algorithms are “hard to break” and are able to work in regimes where all related techniques fail. The second effort targets high-level understanding. While humans are able to quickly grasp both geometric and semantic aspects of a scene, high-level scene understanding remains a challenge for robotics. I present recent work on real-time metric-semantic understanding, which combines robust estimation with deep learning.

The concept of automating vehicles and removing the driver from direct control of the throttle, brake, and steering wheel was first explored nearly 100 years ago. Over the decades since, automation of various features has gradually infiltrated the automobile. Today, on the heels of the DARPA Urban Challenge and Google’s Self-Driving Car Project, we are closer than ever to realizing aspirations of a century ago, but challenges remain. This talk will center on elements of what is known about automation in the vehicle today and our evolution towards self-driving. Topics will include: observations on the use of Advanced Driver Assistance Systems (ADAS) and production level automated driving features (Autopilot, Pilot Assist, Super Cruise, etc.); the shifting nature of what we do in modern vehicles, challenging what is today’s distraction - secondary tasks or driving; and key points to consider regarding the future of robots on our roads. How might the intersection of artificial intelligence embodied in one the most complex activities humans perform - intersect with society’s demand for economical, efficient and safe mobility? How can human factors insight, psychological research, and policy leadership help to accelerate innovations that will someday change how we live and move? How fast might the automated, electrified future of mobility really take hold?

As the technology for autonomous vehicles matures, the broad reach of the technology comes to focus, together with the new challenges and the shifting opportunities. The car that can drive itself under any condition better than any human driver - the holy-grail of autonomous vehicles - may not be as close as once thought. However, it is becoming clear that other opportunities with tremendous economic and social impact may be well within reach. In fact, fielding autonomous vehicles on the ground, in the air, on the water and even in space may transform a number of existing industries and create new ones. In this talk, we discuss three emerging technologies that will allow autonomous vehicles to interact with humans, rapidly react to their environment, and showcase complex autonomy even in miniature form factors, respectively. We also briefly discuss opportunities in business and in teaching of autonomous vehicles.