AI

Joint upper-bound coverage and route-choice utility: an empirical evaluation on two urban proxy tasks

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 08/10/2026, 02:18:24 UTC.

Briefings

This study evaluates whether improved traffic forecast uncertainty coverage enhances route decision quality using offline proxy tasks in Beijing and Chengdu. Results show that while joint upper-bound coverage increased from 83.19% to 92.26% in Beijing M1, it increased lateness by 0.1633 percentage points and travel time by 0.588 seconds. In Chengdu M1, coverage rose from 75.14% to 88.33%, but lateness increased by 0.7848 percentage points and travel time by 3.082 seconds. A separate Chengdu predictor analysis found that reducing speed MAE by 14.91% reduced lateness by 1.4571 percentage points, suggesting coverage alone does not guarantee better outcomes.

Key points

• Joint upper-bound coverage increased from 83.19% to 92.26% in Beijing M1, yet increased lateness by 0.1633 percentage points and travel time by 0.588 seconds.
• Joint upper-bound coverage increased from 75.14% to 88.33% in Chengdu M1, yet increased lateness by 0.7848 percentage points and travel time by 3.082 seconds.

Context

Researchers tested if higher uncertainty coverage improves route decisions using offline proxy tasks with frozen protocols. They used processed road speed data from Beijing and Chengdu to construct tasks with 150 origin-destination pairs, three candidate paths, and 14 test days per city.

Limitations

Results are limited to offline proxy tasks and do not establish online or causal benefits. The study does not account for real-time traffic dynamics or dynamic route adaptation.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 04/11/2026, 02:40:26 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeie2rf25yjellsyffkgiyrprdatzjqqd6wfd62zhzxjyziaczpi5ha