Experiment Recordings

These data sets are recordings of closed-loop experiment runs that demonstrate library functionality. They are recorded as rosbags. All data sets released under CC BY 4.0 license.

e02: Longitudinal Colliding Reciprocal Dance

• Demonstration of colliding reciprocal dance issue in simple longitudinal scenario
• Three mitigation strategies are testing: constraint tightening, conservative deceleration, and none
• Includes RViz config for easy viewing

e01: Short AR slalom with Image Space Potential Field Navigation

• Six second run demostrating ar_isp_field control package
• Guidance control is constant forward throttle and centered steering
• Staggered obstacles obstruct the agent's path
• The controller computes ISP fields from camera input and perturbs the guidance control to avoid collision
• Slideshow for experiment: https://youtu.be/BaHekhZmkfY

Simulation Sequences

These data sets are open-loop simulation sequences provided by Parallel Domain. All data sets released under CC BY 4.0 license.

s05: Hills with transverse traffic

• 175 frame, medium resolution visual of traveling along a hilly road with transverse traffic
• Ground truth, per-pixel semantic segmentation with blur noise
• 1280x720 resolution
• Includes source simulation movies

s04: Road fork

• 300 frame, medium resolution visual of an approach and traversal of a forked road
• Ground truth, per-pixel semantic segmentation with blur noise
• 1280x720 resolution
• Includes source simulation movies

s03: Two car slalom

• 300 frame, medium resolution visual of an approach and slalom between two cars
• Ground truth, per-pixel semantic segmentation with blur noise
• 1280x720 resolution
• Includes source simulation movies

s02: Truck approach from behind

• 215 frame, medium resolution visual of an approach on a truck from behind on an uneven street
• Ground truth, per-pixel semantic segmentation with blur noise
• 1280x720 resolution
• Includes source simulation movies

s01: Low-res fly-through

• 90 frame, low resolution fly through of an empty street with some infrastructure and lightly populated environment
• Ground truth, per-pixel semantic segmentation
• 768x432 resolution
• Includes raw data and rosbag versions