MAN TruckScenes: 740 Scenes of Multimodal Sensor Data for Autonomous Trucking

MAN TruckScenes is a large-scale multimodal dataset for autonomous trucking research, created by MAN Truck and Bus SE. It comprises more than 740 scenes of 20 seconds each, recorded with a heavy truck equipped with 6 lidars, 6 radars, 4 cameras, and a high-precision GNSS. The dataset includes bounding boxes for 27 object classes, 15 attributes, and object tracking to address truck-specific challenges like trailer occlusions.

Use Cases

• Train object detection models based on bounding boxes for 27 object classes.
• Develop sensor fusion algorithms based on synchronized data from 6 lidars, 6 radars, and 4 cameras.
• Study long-range object tracking based on annotations with a range of more than 230 meters.
• Analyze truck-specific perception challenges like trailer occlusions based on the dataset's stated focus.
• Classify driving scenes based on the 34 distinct scene tags mentioned in the description.

Strengths

• Large scale with more than 740 recorded scenes.
• Multimodal sensor suite includes 6 lidars, 6 radars, and 4 cameras.
• Detailed annotations include 27 object classes, 15 attributes, and object tracking throughout scenes.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

MAN Truck and Bus SE

Collection Method

Sensor data recorded with a heavy truck from MAN.

Time Range

null

Freshness

Last updated is unknown.

Geography

null