Yiren Zhao

The Hong Kong University of Science and Technology (Guangzhou)

Unclaimed · AI-structured profileOpenAlex

HKUST (Guangzhou) researcher building risk-aware occupancy for safety-oriented end-to-end autonomous driving.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Insufficient data: this profile has no aggregate output metrics yet. The output axis would measure publication volume, venue distribution, and productivity trends over time; with only two 2026 arXiv preprints in scope, no score is reported.

Impact—

Insufficient data: no citation or influence signals are available. The impact axis would measure citations and downstream uptake of this researcher's work; none exists in the current snapshot, so the axis is left unscored.

Focus0/100

The focus axis scores 0 in this snapshot. The raw field distribution splits evenly — 0.5 to autonomous driving/computer vision and 0.5 to autonomous driving/machine learning — because the two catalogued papers sit on opposite sides of the same application area. With only two papers in scope, this score reflects the thin sample more than scattered interests: the underlying topics, occupancy and end-to-end planning, are tightly aligned.

Collaboration39/100

The collaboration axis scores 39 in this snapshot. Bibliographic records show six distinct coauthors, with repeated partners Jiaxing Chen, Hengduo Zou, Bolin Gao, YuKai Qin, and Lidong Yu — a small, tight-knit team pattern. On both catalogued 2026 papers the researcher sits in a middle author position and is not the corresponding author, indicating a contributor role within the team rather than a lead or supervising position so far.

Activity—

Insufficient data: recent publication cadence cannot be computed. The activity axis would measure output frequency and research continuity over recent years; the snapshot provides no such time-series data, so no activity score is reported.

Reproducibility55/100

The reproducibility axis scores 55, a mean readiness score across two graded papers (reproPapers = 2), both rated medium. For the register-based NAVSIM planner work, headline PDMS gains from 0.937 to 0.951 depend on oracle future-derived risk labels, complicating independent replication; for the safety-oriented occupancy work, GenAD collision reductions of 35-53% rest on small-count open-loop metrics without statistical testing. Moderate readiness with clear caveats.

Research summary

Yiren Zhao is an early-stage researcher at The Hong Kong University of Science and Technology (Guangzhou) whose Evolve catalog to date centers on safety-oriented end-to-end autonomous driving. Two 2026 arXiv preprints define the profile, both introducing risk-aware occupancy as a representation for driving policy: one couples it to a register-based planner on the NAVSIM benchmark, reporting PDMS improving from 0.937 to 0.951, while the other evaluates the same idea on nuScenes and GenAD, where collision rates drop by 35-53% in open-loop settings. Evolve's review notes temper both results: the NAVSIM gains rely on oracle future-derived risk labels, and the small-count open-loop collision metrics lack statistical testing. The researcher appears in middle author positions on both papers and is not corresponding author, working within a compact circle of six distinct coauthors including Jiaxing Chen, Hengduo Zou, Bolin Gao, YuKai Qin, and Lidong Yu. Field assignment splits evenly between autonomous driving/computer vision and autonomous driving/machine learning. Reproducibility signals are moderate — a mean readiness score of 55 across the two graded papers, both rated medium. Output, impact, and activity metrics are not yet available, so this profile should be read as an early snapshot rather than a mature record.

Collaboration pattern

Collaboration is compact and recurring: six distinct coauthors appear across the two catalogued 2026 papers, with Jiaxing Chen, Hengduo Zou, Bolin Gao, YuKai Qin, and Lidong Yu as the frequent partners. The researcher holds middle author positions and is not corresponding author on either paper, consistent with a contributor inside a small, stable team at HKUST (Guangzhou) rather than an independent lead. The collaboration axis scores 39 on this basis.

Research keywords

End-to-end autonomous drivingRisk-aware occupancyTrajectory planningMultimodal trajectory predictionOccupancy predictionBird's-eye view representationNAVSIM benchmarkPDMS planning metricCollision rate evaluationnuScenes dataset

Highlights

Risk-aware occupancy lifts a register-based NAVSIM planner from PDMS 0.937 to 0.951
In the 2026 register-planner preprint, risk-aware occupancy is added to an end-to-end planner on the NAVSIM benchmark, raising PDMS from 0.937 to 0.951. Evolve's verdict cautions that the headline gains rely on oracle future-derived risk labels, so the improvement should be read with that dependency in mind.
GenAD collision rates drop 35-53% with risk-aware occupancy
The safety-oriented end-to-end driving preprint reports collision-rate reductions of 35-53% on GenAD when risk-aware occupancy guides the policy. Evolve notes these open-loop results involve small collision counts and lack statistical testing, so the safety claim is promising but not yet statistically established.
One coherent methodological thread across two 2026 preprints
Both catalogued papers advance the same idea — risk-aware occupancy as a safety signal for end-to-end driving — one on NAVSIM with a register-based planner, the other on nuScenes/GenAD, showing a consistent early research agenda spanning machine-learning and computer-vision framings.

Timeline

2026First Evolve-catalogued preprint proposes risk-aware occupancy for safety-oriented end-to-end autonomous driving, evaluated on nuScenes and GenAD.
2026Second catalogued preprint, RRDrive, integrates risk-aware occupancy into a register-based NAVSIM planner, lifting PDMS from 0.937 to 0.951.
2026Both 2026 papers receive medium reproducibility readiness (mean score 55), with the researcher in middle author positions within a six-person coauthor circle.

Papers in the Evolve catalog

Frequent co-authors

Jiaxing Chen×2 Hengduo Zou×2 Bolin Gao×2 Yiren Zhao×1 YuKai Qin×1 Lidong Yu×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Radar scores are derived from OpenAlex-based bibliometric data limited to the two papers in Evolve's catalog.
Output, impact, and activity axes lack underlying data in this snapshot and are intentionally left unscored.
Both catalogued papers are 2026 arXiv preprints; peer-review status is not captured in this data.
No ORCID or country information was provided; the profile is name-resolved from public bibliographic data, and 'Yiren Zhao' is a relatively common name.