Tsinghua researcher building risk-aware occupancy representations for safety-oriented end-to-end autonomous driving.
2
papers here
2
reviewed by Evolve
2
first-author
0
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—
No output metrics are available yet — the catalog holds only two 2026 arXiv preprints, with no publication counts or venue history. This axis would measure the volume, cadence, and consistency of published output across years and venues.
Impact—
No impact signals are available — citation counts, h-index, and downstream adoption are absent from the input data. This axis would measure how often the two risk-aware occupancy preprints are cited or built upon by the driving community.
Focus0/100
Focus scores 0 in the current radar. The raw field distribution splits evenly — 50% autonomous driving / computer vision and 50% autonomous driving / machine learning — so the cataloged work spans two sub-field labels, though both orbit one problem, risk-aware occupancy for end-to-end driving. With only two papers in the sample, this score is noisy and should be read cautiously.
Collaboration39/100
Collaboration scores 39, based on 6 distinct coauthors across the catalog. Both papers are first-authored (non-corresponding), and the recurring pool is small — Hengduo Zou, Yiren Zhao, Bolin Gao, YuKai Qin, and Lidong Yu — suggesting a compact, stable team rather than a wide external network.
Activity—
Activity data is insufficient: the snapshot provides no publication-frequency or per-year volume series, so recent research tempo cannot be scored. This axis would measure publishing cadence over recent years, here inferable only from the two 2026 preprints.
Reproducibility55/100
Reproducibility scores 55, the mean over 2 papers assessed, both rated medium reproReadiness. Evolve's verdicts flag concrete gaps: the RRDrive PDMS gain (0.937→0.951) depends on oracle future-derived risk labels, and the 35-53% GenAD collision reductions come from small open-loop counts without statistical testing — both fixable limitations.
Research summary
Jiaxing Chen is a researcher at Tsinghua University working on safety-oriented end-to-end autonomous driving, with a catalog profile built from two first-author arXiv preprints, both dated 2026. The work centers on risk-aware occupancy: representing the scene so that a planner can reason about where risk concentrates rather than only predicting trajectories. The first paper proposes the representation for end-to-end driving, reporting GenAD collision-rate reductions of 35-53% alongside nuScenes evaluation, though Evolve notes these open-loop counts are small and statistically untested. The follow-up, 'Driving on Registers, Reasoning on Risk,' integrates risk-aware occupancy into a NAVSIM register-based planner, lifting PDMS from 0.937 to 0.951, with headline gains relying on oracle future-derived risk labels. The field distribution splits evenly between the computer-vision and machine-learning sides of autonomous driving, and the papers span occupancy prediction, bird's-eye view representation, trajectory planning, and multimodal trajectory prediction. Reproducibility readiness is rated medium for both papers, with a mean score of 55 across the two assessments. Collaboration is currently small-scale — six distinct coauthors including Hengduo Zou, Yiren Zhao, Bolin Gao, YuKai Qin, and Lidong Yu — while the output, impact, and activity axes remain unscored for lack of data.
Collaboration pattern
The cataloged work is a compact-team effort: 6 distinct coauthors, with Hengduo Zou, Yiren Zhao, Bolin Gao, YuKai Qin, and Lidong Yu recurring across papers, and Jiaxing Chen first author on both (non-corresponding). The collaboration score of 39 reflects a small, stable core group rather than a broad external network; no industry-lab or cross-institution patterns are visible in the current snapshot.
Research keywords
Highlights
PDMS lifted from 0.937 to 0.951 on NAVSIM via risk-aware occupancy in a register-based planner
Collision-rate drops of 35-53% on GenAD from a safety-oriented occupancy representation
First-author leadership on both cataloged 2026 preprints
Both Evolve-reviewed papers list Jiaxing Chen as first author (non-corresponding), released on arXiv in 2026, indicating early ownership of the risk-aware occupancy line of work within a six-person collaborator pool.
Timeline
- 2026Publishes the first-author preprint 'Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous Driving' on arXiv, proposing risk-aware occupancy with bird's-eye view representation, evaluated on GenAD and nuScenes.
- 2026Follows with the first-author preprint 'Driving on Registers, Reasoning on Risk,' integrating risk-aware occupancy into a NAVSIM register-based planner and lifting PDMS from 0.937 to 0.951.
- 2026Both preprints enter the Evolve catalog with medium reproducibility readiness, consolidating an early, safety-focused agenda built with six distinct coauthors.
Papers in the Evolve catalog
Frequent co-authors
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 computed from bibliometric data only; the output, impact, and activity axes lack sufficient data and are deliberately unscored.
- The profile rests on two 2026 arXiv preprints; no ORCID, country, or peer-reviewed venue information is present, so career facts cannot be verified.
- 'Jiaxing Chen' is a common name; this profile is name-resolved from public bibliographic data and the two reviewed papers, and identity should be confirmed via ORCID when available.
- Benchmark gains carry methodological caveats from Evolve's review: reliance on oracle future-derived risk labels and small-count open-loop metrics without statistical testing.