SPIDER: Simultaneous Pickup-Delivery Instances for Demand-controlled Emission-aware Routing

SPIDER (Simultaneous Pickup-Delivery Instances for Demand-controlled Emission-aware Routing) is a benchmark dataset of 1,152 vehicle routing instances with simultaneous pickup and delivery (VRPSPD). The instances are generated from a complete factorial design over five controlled dimensions: customer size (50, 100, 200, 400, 600, 1000), spatial layout (random, clustered, mixed), pickup-to-delivery demand ratio (low, moderate, high, extreme), vehicle-capacity tightness (very tight, tight, medium, slack), and time-window regime (open, narrow, medium, wide). Crossing all five dimensions yields 6 x 3 x 4 x 4 x 4 = 1,152 instances, each appearing exactly once. Customers are placed on a 200 x 200 integer grid with the depot fixed at the centre. Each instance specifies depot and customer coordinates, delivery and pickup demands, service times, time-window bounds, and an optional block of load-dependent emission parameters. A master index (index.csv) provides one row per instance with the encoded design factors and derived quantities for filtering and grouped analysis. Reader and distance-matrix utilities are included. The dataset supports distance, capacity, time-window, and emission-aware objective formulations, and is suitable for exact methods, heuristics, metaheuristics, matheuristics, and learning-based solvers. All 1,152 instances passed feasibility verification.