2026-10-10

7 arXiv categories· 96 journals· 3432 candidates — 6 worth your time· 53 without an open abstract

Headline

Risk-aware maneuver planning for autonomous vehicles under blind-spot occlusions

Today, together

tag shift
no tag ran above its 30-day average
shares references
no items share references today counted over the 6 papers published above
on the same shoulders
no foundation work was cited by four of the day's papers counted over the 55 of this day's 59 named papers whose reference lists we have — standing on the same older work is not the same as being about the same thing
first appearance
South China Normal University; University of Minnesota; Wrocław University of Science and Technology none of them on a paper we published in the 120 days before this issue

Two counts, not one counted twice: every row here but “on the same shoulders” counts the papers published above, and that one counts every paper this day named — those and the ones under Also published today, whose abstracts we could not read.

DREAM introduces a defensive risk-aware maneuver planning framework for autonomous vehicles in heterogeneous traffic using partial differential equations and model predictive control with control barrier functions. Monthly Rhythms of Intercity Travel analyzes the 2023 seasonal dynamics and temporal fluctuations of population mobility across Chinese cities using AMap location-based migration data. LEON presents a self-supervised representation learning framework that generates urban location embeddings from OpenStreetMap amenity distributions mapped over H3 hexagonal grids using Graph Masked Autoencoders.

codedatapublished: Transportation Research Part C Emerging Technologies

Zian Wang, Yiming Shu, Zejian Deng, et al.
Transportation Research Part C Emerging Technologies

University of Hong Kong

Autonomous Vehicle Technology and Safety · Robotic Path Planning Algorithms · Traffic control and management · University of Hong Kong · Tsinghua University

DREAM: Defensive risk-aware enhanced maneuver planning for autonomous vehicles in heterogeneous traffic

The DREAM framework models occluded vehicular hazards using an advection-diffusion-reaction partial differential equation and integrates the generated risk field into Frenet-domain model predictive control via control barrier functions. Evaluated across CARLA simulations, ROS 2 hardware tests, and the inD dataset, the system aborted unsafe merges approximately one second prior to line-of-sight recovery, improved minimum clearance to hidden agents by 0.428 meters, and decreased collision or near-miss events from 29/30 to 19/30 in high-density scenarios.

Why it matters — This approach enables autonomous planners to proactively hedge against hidden dynamic hazards behind large vehicles without resorting to overly cautious slowdowns near unthreatened occlusions.

Transportation Research Part C Emerging Technologies · doi · code

published: Applied Spatial Analysis and Policy

Fan Xiao, Xueying Mai, Meicheng Xiong
Applied Spatial Analysis and Policy

South China Normal University

Human Mobility and Location-Based Analysis · Transportation Planning and Optimization · Regional Economics and Spatial Analysis · mobile phone data · China · South China Normal University

Monthly Rhythms of Intercity Travel: A Seasonal Analysis of Population Mobility Between Chinese Cities

Using AMap location-based migration data across 2023, this study analyzes the monthly dynamics and temporal fluctuations of population movement across Chinese cities at national, municipal, and city-pair scales. Applying K-means clustering and temporal Gini coefficients, the analysis classifies urban travel rhythms into four distinct types and finds that national flows peak in January, August, and October. Economically prominent cities demonstrate higher total mobility with lower seasonal variability, whereas travel between paired cities exhibits the highest stability when restricted to intra-provincial or metropolitan boundaries.

Why it matters — It quantifies nation-wide seasonal baseline fluctuations in intercity travel across an entire year, establishing that economic strength and geographic proximity dampen temporal swings in urban mobility.

Caveat: The findings rely entirely on mobile location data from a single provider during 2023, a year that may reflect post-pandemic recovery anomalies in China's travel behavior.

Applied Spatial Analysis and Policy · doi

preprint

Szymon Soltysiak, Radoslaw Malek, Jedrzej Kusnierz, et al.
arXiv cs.LG

Advanced Graph Neural Networks · Geographic Information Systems Studies · Urban Design and Spatial Analysis · self-supervised learning · graph neural network · satellite imagery

LEON: Location Embeddings from OSM Neighborhoods via Hexagonal Graph Masked Autoencoders

LEON is a self-supervised representation learning framework that generates location embeddings by applying Graph Masked Autoencoders to OpenStreetMap amenity distributions mapped over H3 hexagonal grids. Evaluation against benchmark datasets yielded up to a 1.87% accuracy increase in EuroSAT satellite image classification, alongside improvements in predicting housing prices, crime rates, and urban analytics metrics.

Why it matters — This approach demonstrates that structured, generalizable spatial representations can be learned directly from globally available, crowdsourced vector data without relying on labeled targets or imagery.

arXiv · pdf · doi

preprint

Domingos Barbosa, Dalmiro Maia, Bruno Morgado, et al.
arXiv astro-ph.IM

satellite imagery

Assessment of sky brightness in central Portugal: the Dark Sky destination "Aldeias do Xisto" case study

Using VIIRS satellite data, MODIS cloud cover estimates, and GIS spatial modeling from 2012 to 2017, the authors mapped artificial light at night across 1,624 km² of Portugal's Schist Villages region. Ground validation via zenith Sky Quality Meter photometers confirmed dark sky locations, recording a maximum night sky brightness of 21.4 mag/arcsec² in the darkest area.

Why it matters — This provides an accessible, low-cost framework combining remote sensing and field photometers to evaluate rural sites for dark sky preserves and astrotourism prior to heavy infrastructure investment.

arXiv · pdf

preprint

Megh Marathe, K Ranade, L. Ramakrishnan, et al.
arXiv cs.HC

interview

Redistributing Harm: Document Transition and the Limits of Trans Inclusion in India's Identity Systems

Qualitative focus groups conducted in mid-2025 with 16 trans and gender-diverse participants in India reveal that updating official identity documents entails navigating unclear protocols, inconsistent evidentiary requirements, and mistreatment by officials. Discrepancies between official records cause severe disruptions across employment, housing, civic participation, education, and access to justice, imposing financial penalties such as excess tax withholdings and job insecurity. Obtaining gender-concordant documents often redistributes administrative burden, as individuals recognized as male forfeit affirmative action benefits, welfare provisions, and specialized legal protections while continuing to face social discrimination.

Why it matters — The study demonstrates that administrative systems designed to include trans populations can produce secondary harms, showing that technical legibility within existing legal structures does not eliminate structural inequality and may worsen economic vulnerability.

Caveat: The findings rely on a small sample of 16 participants composed predominantly of transmasculine and gender-diverse individuals, which may not capture the full range of administrative outcomes across all gender identities in India.

arXiv · pdf

preprint

Nazmul Haque, Md Asif Raihan. Md. Hadiuzzaman
arXiv cs.LG

regression · recurrent neural network

Neuro-Memory Fuzzy Inference System for Mimicking Human-like Car Following Behavior

The Neuro-Memory Fuzzy Inference System (NeMeFIS) models human car-following action by separating acceleration and deceleration into five cognitive memory structures across 54 setups on Arterial, Collector, and Rural Highway corridors. Across these configurations, the model outperformed Linear Regression, ANFIS, and LSTM benchmarks in simulating real driving responses. Fuzzy rule parsing demonstrated that procedural memory primarily governs acceleration decisions, whereas declarative and semantic memory regulate deceleration due to higher risk perception and cognitive load.

Why it matters — By assigning different cognitive memory frameworks to distinct driving maneuvers, this approach allows autonomous vehicle algorithms to mimic human reaction thresholds and braking choices more accurately than standard neural architectures.

arXiv · pdf

Still cited

Mimí Sheller, John Urry (2006), The New Mobilities Paradigm 1 of today's items cite it · 30 of 3240 in the archive stand on it

Also published today

These appeared today in journals we track. Their abstracts are not openly available, so we cannot summarise them.