Society | HXI - Human-centered eXtended Intelligence

CLARO: Turning Complexity into Clarity

Mon, 14 Sep 2026 00:00:00 +0000

Overview

Challenges like climate, housing, health, and neighborhood revitalization are deeply interconnected, and the tools used to reason about them are not. Dashboards, static reports, and siloed studies produce disjointed analysis and contradictory recommendations, and qualitative evidence rarely sits alongside the quantitative. Decision-making also tends to exclude the perspectives most affected by the outcome.

CLARO (Collaborative Lab for Analytics, Reality-modeling, and Observation) is a platform for building interactive simulations that make advanced data modeling and scenario planning usable by policymakers, planners, and communities. In Spanish, claro means clear.

The platform is organized as four layers: data ingestion, modeling, simulation, and visualization. Structured and unstructured sources (text, audio, video) feed a modeling layer built on knowledge graphs, large language models, and retrieval-augmented generation. Above it sit digital twin, GIS, and game-engine simulation, and a visualization layer spanning 2D, 3D, voice, and VR/AR/XR. A conversational interface lets a user pose a question in natural language and receive generated maps, charts, and scenarios.

The design goal is a shared view: the same underlying data, rendered through whichever lens a given stakeholder works in, so that technical expertise is not a precondition for participating in the decision.

CLARO formalizes an intersection the Design Lab has worked in for years, bringing together humanity-centered design, systems and behavioral modeling, and computing power.

Current use case

San Diego affordable housing. CLARO is being applied to help policymakers and developers identify and assess land opportunity, using live data from the California Housing Partnership (CHPD) and SANDAG.

Funding and External Collaborations

CLARO is a strategic initiative of The Design Lab at UC San Diego, developed in collaboration with the San Diego Supercomputer Center and its Artificial Intelligence and Data Innovation Lab (ADIL), which provides data ingestion, storage, and API infrastructure. Nadir Weibel serves as advisor and co-leads the interface and visualization and conversational AI workstreams. NVIDIA platforms, including Omniverse and Video Search and Summarization, are under evaluation for immersive visualization and urban sensing.

Designing Smart and Autonomous Vehicles

Thu, 25 May 2023 00:00:00 +0000

Overview

Despite huge advancements in Autonomous Vehicle (AV) technology, real-world adoption is hindered by a lack of calibrated passenger trust in vehicle decisions. In this research agenda, we combine Virtual Reality (VR), biometric measurement, human-centered design, and data science methods to uncover factors impacting trust with Autonomous Vehicles (AVs) at the level of the individual as opposed to taking a one-size-fits-all approach.

We aim to establish a generalizable theory connecting personal characteristics to AV communications in realistic driving contexts as well as underpin why certain designs are more effective than others at eliciting calibrated trust, improving driving outcomes through enhanced situational awareness, and alleviating stress. This research agenda will inform future driver-AV interactions and impact important design decisions for the automotive and self-driving car industry in the years ahead.

Calibrating Trust in Autonomous Vehicles

Trust in an AV is not uniform across passengers, so we study it at the level of the individual. Surveying 1,457 young adults across psychosocial traits, cognitive attributes, driving style, prior experience, and perceived risks and benefits, and modeling the results with machine learning and SHAP, we find that perceptions of AV risks and benefits, attitudes toward feasibility and usability, institutional trust, prior experience, and mental models predict trust far better than psychosocial traits or driving style do.

How the vehicle communicates matters just as much as who is listening. In a simulated driving study with 232 participants, errors in an AV’s explanations reduced comfort in relying on the vehicle, confidence in its ability, and explanation satisfaction, even though the driving itself was identical. Perceived harm and driving difficulty amplified the damage, so the contexts where an explanation matters most are also the ones where getting it wrong costs most.

Together these results argue against one-size-fits-all AV communication, and toward explanations matched to the passenger and the situation.

Simulating and Sensing the Driver

Studying any of this requires being able to observe a driver closely without putting anyone on a real road. Head-mounted VR makes that possible, but existing simulators tend to track only eye movement, and their bulky outside-in rigs and Unreal-based architectures put them out of reach for most interaction researchers.

DriveSimQuest is our answer: a VR driving simulator and research platform built on the Meta Quest Pro and Unity, capturing gaze, facial expression, hand activity, and full-body gesture in real time. It is deliberately easy to deploy, so that studying drivers' affective states becomes a matter of designing the study rather than building the rig.

That platform is what lets the rest of this work ask sharper questions, about what a driver is attending to, what they are feeling, and when a system should intervene.

Coaching Human Drivers at Fleet Scale

The same safety-critical setting looks different when the driver is a person rather than a system. Fleet drivers are involved in collisions that impose severe financial costs and endanger lives, and fleet companies lean on one-to-one coaching to prepare them. Yet the people, practices, and technologies that shape that coaching have remained largely unexamined.

We characterize fleet coaching from both sides, surveying coaches about current practice and interviewing drivers about how coaching actually lands. Manager-led coaching outperforms self-coaching across experiential outcomes, and the reasons why point toward what would have to be true for coaching to scale without losing the relationship that makes it work.

Funding and External Collaborations

The autonomous vehicle trust work is a collaboration between the UCSD departments of Cognitive Science, Computer Science and Engineering, and external industry partners. The fleet driver coaching research is conducted with Lytx.

Publications

DriveSimQuest: A VR Driving Simulator and Research Platform on Meta Quest with Unity

Nishanth Chidambaram, Weichen Liu, Manas Bedmutha, Nadir Weibel, Chen Chen

Comparative Effectiveness of Coaching Modalities in Commercial Fleet Operations

Nadir Weibel, Matin Yarmand, Cole Biehle, Elysia Mac, Huimeng Lu, Vivian Xiang, Gabriella Strudler, Pari Hathiram

Evaluating and Optimizing Coaching Methodologies for Fleet Safety and Performance: An Evidence-Based Analysis of Differentiation and Optimization Opportunities

Nadir Weibel, Matin Yarmand, Cole Biehle, Elysia Mac, Huimeng Lu, Vivian Xiang

Predicting trust in autonomous vehicles: Modeling young adult psychosocial traits, risk-benefit attitudes, and driving factors with machine learning

Robert A. Kaufman, Emi Lee, Manas Bedmutha, David Kirsh, Nadir Weibel

What did my car say? impact of autonomous vehicle explanation errors and driving context on comfort, reliance, satisfaction, and driving confidence

Robert A. Kaufman, Aaron Broukhim, David Kirsh, Nadir Weibel

RECODE|Health: Research Center for Optimal Digital Ethics

Thu, 23 Sep 2021 00:00:00 +0000

Overview

ReCODE Health is here to support technologists, researchers, ethicists, regulators, institutions and participants involved in the digital health research process. Our value proposition is to increase awareness of ethical principles and practices from the earliest stages of technology design to the deployment of digital health research.

For Technology Creators – We offer assistance in developing Terms of Service and End User Licensing Agreements that don’t conflict with Federal Regulations for protecting research participants. Researchers can then use your measurement product in digital health research.

For Researchers
We provide protocol development consultation, education and decision-making tools.

For Participants
We offer checklists to help you make decisions about participating in digital health research.

For Institutions
We conduct risk analysis assessments with a goal of increasing awareness of the unknown unknowns that are inherent in digital health research.

For Ethicists
We work with you to apply ethical principles to digital health research.

More Info here: http://recode.health

Funding and External Collaborations

Our ReCODE Health team is composed of faculty and students affiliated with the UC San Diego Schools of Medicine, Engineering, and Humanities and Social Sciences as well as the Institute of Practical Ethics, Research Ethics Program, Qualcomm Institute, Calit2, Center for Wireless and Population Health Systems, Design Lab, Center for Wireless Sensors, Connected and Open Research Ethics, Human Research Protection Program (HRPP), and the Center for Health Promotion and Equity.

ReCODE Health is supported by University of California San Diego, and its programs have been funded by the Robert Wood Johnson Foundation, the National Institute of Drug Abuse at NIH, the National Science Foundation, and IBM Research.