Spatial perception has witnessed an unprecedented progress in the last decade. Robots 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) or can structure the environment to simplify perception (e.g., industrial robotics), safety-critical applications of robotics in the wild, ranging from self-driving vehicles to search & rescue, 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 certifiably robust spatial 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 metric-semantic 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. I discuss these efforts and their applications to a variety of perception problems, including mesh registration, image-based object localization, and robot Simultaneous Localization and Mapping.
Moderator: David Keith Panelists (4-minute statement each): Kent Larson Carlo Ratti Sarah Williams Jinhua Zhao
Given the severe mobility challenges in urbanizing areas, numerous visions for designing urban mobility systems are discussed by policymakers, planners, and industry. These visions must anticipate technological and sociodemographic developments, while accounting for the constraints of operator business models and environmental concerns. In this session, MIT faculty will share and discuss their ideas for urban mobility systems around the globe, considering both promising technologies as well as heterogeneities among the world’s urban centers.
Moderator: Randall Field Speakers: - Matthias Winkenbach (14 minutes) "Opportunities and Challenges of Urban Delivery of Goods" - John Hansman (14 minutes) "Opportunities and challenges for urban air mobility"
Urban areas around the globe face increasing mobility challenges. Demand for both passenger and freight services continue to increase, straining already congested systems. Opportunities to build new infrastructure to address these challenges are limited. Therefore, novel system designs are needed to support mobility in future urban environments. For passenger transportation, Urban Air Mobility systems could create additional capacity through largely decoupling transportation from the confinements of the ground. For freight transportation, existing ground infrastructure (e.g. metro systems) could be leveraged systematically and autonomous systems in combination with additive manufacturing techniques for localized production could disrupt urban logistics.
Moderator: Randall Field Speakers: - Sertac Karaman (17 minutes) "The path towards autonomous vehicles on our roads" - Sanjay Sarma (17 minutes) "Digitalization of the mobility value chain: opportunities and implications"
Over the past decade, new digital technologies have re-defined mobility in urban areas through new on-demand mobility services offered through a sharing economy model. This session will explore future opportunities associated with digital transformations of the mobility value chain and will assess the implications linked to these transformations from a strategic perspective. In particular, the technological foundations of in-vehicle digitalization will be explored for the case of autonomous vehicles, with a focus on assessing current technical implementations and potential technical solutions.
Moderator: Joanna Moody Speakers: - Sergey Paltsev (9 minuets) "Mobility policy, energy demand, and global scenarios" - Jessika Trancik (9 minuets) "Low-carbon mobility technology development: Measuring progress and predicting innovation opportunities using new models"
Mobility systems are constantly changing. Currently, the availability of on-demand mobility services and the development of new vehicle technologies (e.g. electric vehicles) is altering the way we travel in urban areas. At the same time, policymakers around the world tackle the environmental challenges associated with mobility systems through new policies including emission standards and driving restrictions. This session sets out to (i) provide an overview of emerging vehicle technologies for passenger ground transportation, especially with regard to fuels and powertrains, (ii) outline the interaction of technology adoption with different policy scenarios, and (iii) describe current adoption of new technologies and future innovation opportunities.
The MIT Integrated Learning Initiative (MITili) is a cross-disciplinary, Institute wide initiative to foster quantitative and rigorous research about how people learn and how knowledge from that research can enhance learning from school through adult professional education. MITili aims to integrate knowledge from psychology, economics, neuroscience, engineering, and public policy in pursuit of these goals. The work of the future will require life-long learning, and knowledge from learning science ought to enhance that learning. I will show how knowledge from learning science can enhance work-place learning. I will also review how technology might enable, and in some cases disable, learning. I will also share recent findings about how sleep matters for higher education. Finally, I will share some evidence about the brain bases of adult learning.
Machine learning has made tremendous progress over the last decade. It's thus tempting to believe that ML techniques are a "silver bullet", capable of making progress on any real-world problem they are applied to.
But is that really so?
In this talk, I will discuss a major challenge in the real-world deployment of ML: making ML solutions robust, reliable and secure. In particular, I will survey the widespread vulnerabilities of state-of-the-art ML models to various forms of noise, and then outline promising approach to alleviating these deficiencies as well as to making models be more human-aligned.
Computing near the sensor is preferred over the cloud due to privacy and/or latency concerns for a wide range of applications including robotics/drones, self-driving cars, smart Internet of Things, and portable/wearable electronics. However, at the sensor there are often stringent constraints on energy consumption and cost in addition to the throughput and accuracy requirements of the application. In this talk, we will describe how joint algorithm and hardware design can be used to reduce energy consumption while delivering real-time and robust performance for applications including deep learning, computer vision, autonomous navigation/exploration and video/image processing. We will show how energy-efficient techniques that exploit correlation and sparsity to reduce compute, data movement and storage costs can be applied to various tasks including image classification, depth estimation, super-resolution, localization and mapping.
Why do people believe and share misinformation, including entirely fabricated news headlines (“fake news”) and biased or misleading coverage of actual events ("hyper-partisan" content)? The dominant narrative in the media and among academics is that we believe misinformation because we want to – that is, we engage in motivated reasoning, using our cognitive capacities to convince ourselves of the truth of statements that align with our political ideology rather than to undercover the truth. In a series of survey experiments using American participants, my colleagues and I challenge this account. We consistently find that engaging in more reasoning makes one better able to identify false or biased headlines - even for headlines that align with individuals’ political ideology. These findings suggest that susceptibility to misinformation is driven more by mental laziness and lack of reasoning than it is by partisan bias hijacking the reasoning process. We then build on this observation to examine interventions to fight the spread of misinformation. We find - given this smaller-than-believed role of partisan bias - that crowdsourcing can actually be a quite effective approach for identifying misleading news outlets and news content. We also demonstrate the power of making the concept of accuracy top-of-mind, thereby increasing the likelihood that people think about the accuracy of headlines before they decide whether to share them online. Our results suggest that reasoning is not held hostage by partisan bias, but that instead our participants do have the ability to tell fake or inaccurate from real - if they bother to pay attention. Our findings also suggest simple, cost-effective behavioral interventions to fight the spread of misinformation.