DARPA CIDAR Challenge
Passive imaging.
Long-range distance.
Estimate the distance to a target from images taken kilometers away, without emitting light or other signals toward it. The challenge combines demanding accuracy requirements with incomplete information about the scene and imaging conditions.
Challenge setting
2.5–10 km
Passive observation
Metric distance
Images in.
Physical distance out.
- Available data
- Visible and infrared images at different focal lengths, event-camera sequences, and camera specifications including focal length, field of view, and resolution.
- Required output
- A numerical distance estimate for each target, expressed in physical units.
- Challenge scale
- The published scoring criteria cover 2.5, 5, 7.5, and 10 km. The stated challenge goal is accuracy of ±5 m at 10 km or more, with low computational latency.
Rich imagery.
Incomplete information.
The cameras do not share exactly the same viewpoint. Images from different modalities can contain slight parallax, while precise relative camera positions are unavailable.
Scene and measurement conditions are incompletely characterized. The information available during the challenge did not fully specify target material properties, spectral response, or environmental conditions.
Reference distances were unavailable to participants. Teams submitted predictions, but the feedback available to us did not include per-target reference values or errors.
Distance sensing
without active illumination.
Passive ranging is relevant to perception systems that must observe distant targets without illuminating them. The challenge asks whether image information can support accurate physical measurements at long range, under conditions where direct measurement and complete scene information are difficult to obtain.