Zhi-Qi Cheng, Ph.D.

Assistant Professor of Computer Science & Systems

Multimodal Generative AI · Embodied Intelligence · Intelligent Transportation

Ex‑CMU, Google, Microsoft · Intel Ph.D. Fellowship · IBM Outstanding Student Scholarship

  • 📧 Email: zhiqics@uw.edu
  • ☎︎ Phone:  412‑623‑9121
  • 📍 Office: Milgard Hall 221.6

Sponsored Projects (selected)
Select sponsored projects highlighting real‑world deployments of multimodal AI in transportation, public safety, manufacturing, and national security.
Funding: DARPA KAIROS DARPA AIDA IARPA DIVA NIST PSIAP USDOT UTC Mobility21 CMU MFI
Funding update. The listed awards (DARPA, IARPA, NIST, USDOT Mobility21, CMU MFI) were made during my CMU tenure. At the University of Washington, I am pursuing new funding to sustain and expand these research directions, with emphasis on student support, computational infrastructure, and data acquisition. Interested collaborators and sponsors are warmly invited to connect.
DARPA Logo

DARPA KAIROS

Schema‑guided event understanding from video, image & text streams (Tech‑lead, 2019 – 2024). (final system description)

Mobility 21 UTC – sensing and perception pipeline schematic

Mobility 21 UTC

Semantic perception module for AVs to detect & predict road‑user behaviour (2022 – 2023) (final report).

Digital Twins for Manufacturing Resiliency – supply‑chain twin illustration

Digital Twins for Manufacturing Resiliency

ML‑driven twin platform for forecasting material‑supply shocks via multimedia analytics. The system fuses multi‑modal demand/supply signals (news, logistics feeds, ERP exports) with temporal models (sequence learners & temporal GNNs) and a supply‑chain knowledge graph linking tier‑n suppliers. Outputs include early‑warning dashboards, risk propagation maps, and what‑if stress tests for procurement. (project page; arXiv preprint).

HA‑VLN Overview

CMU–AIST Bridge Project

Human‑Inclusive Dynamic Control & Navigation (CMU–AIST Bridge, 2024 – present). Key publications: HA‑VLN (NeurIPS 24 Spotlight); ProMQA (NAACL 25 Oral); VG Dialogue (AAAI 25 Oral).

NIST PSIAP – real-time public safety analytics (smartphone video localization example)

NIST PSIAP 2017

Geo-spatial fusion of live & crowd-sourced video for first-responder dashboards (2017 – 2019).

IARPA DIVA – multi-camera activity analytics example

IARPA DIVA

Real-time multi-camera analytics with graph reasoning for critical-activity detection (2017 – 2021).