Perception Uncertainty and Downstream Planning Risk in Autonomous Electric Taxis: A Layered Evaluation Framework

Title: Perception Uncertainty and Downstream Planning Risk in Autonomous Electric Taxis: A Layered Evaluation FrameworkSimulator: CARLA 0.9.15 (Town10HD_Opt map)Platform: Windows Description:This dataset contains simulation outputs supporting the paper"Perception Bias Propagation in Autonomous Electric Taxis."Experiments cover 4 environmental conditions (E1-E4) and 2 pedestrianclusters (A/B), with 100 runs per condition (800 total).Raw per-episode metrics and aggregated summary statistics are includedalong with all configuration scripts required to reproduce the experiments. Files:- raw_results.csv: per-episode metrics for all 800 runs- statistical_output.txt: t-tests, p-values, Cohen's d, effect sizes- cvar_results.csv: CVaR-adjusted collision probability results per condition/cluster- cvar_summary.xlsx: CVaR-adjusted deceleration and abort rate summaries- cvar_report.txt: CVaR methodology note and full adjusted results report- cvar_planner.py: CVaR post-processing script- run_experiment.py: main experiment script- analyze_results.py: statistical analysis script- sensor_config.py: sensor configuration- weather_conditions.py: weather condition definitions- pedestrian_config.py: pedestrian cluster definitions