Outdoor Long Range Object Detection Experiments with Event-Based Sensors

An event-based sensor (EBS) is an imaging device that detects changes in local irradiance on the focal plane array (FPA) by using a readout integrated circuit (ROIC) with a thresholding operation. These sensors encode local irradiance changes on the FPA as events registered by the device, then read out these events and their time of occurrence in a sparse digital output format. These devices are interesting for various remote sensing applications because the sparse output can translate to a ~100x bandwidth improvement relative to traditional frame-based sensors (FBS). This bandwidth improvement may lead to a ~100x latency improvement and ~100x lower power and cooling requirement for an electro-optical (EO) subsystem. EBS's bandwidth advantage can be leveraged to increase the imaging trade space between spatial resolution, temporal resolution, and field of view, enabling a more effective imaging system for some applications. Given the current maturity of the technology, EBS holds potential to improve imaging capabilities in areas such as autonomous object detection and localization while relaxing processing requirements for overall system design architectures. In principle, EBS sensors provide opportunities for more efficient image processing algorithms due to reduced data output from the sensor's FPA. However, one limitation of EBS technology is its performance in contrast-limited scenarios, which can be attributed to the necessity of local log irradiance changes meeting a certain threshold prior to registering as events which can be read out to the processor. In some challenging imaging scenarios this can result in missed data, where an object may be present though the sensor is not registering events, resulting in false negative detection decisions made by the system. This work aims to better characterize this tradeoff through collecting imagery data in a real-world outdoor setting that can be used to assess the merits of the technology relative to existing alternatives, such as FBS. Specifically, a series of tests will be conducted imaging hot air balloons from (>1km) as they rise from the ground and translate across the field of view (FoV) of the EBS sensor. This raw data will then be post-processed using algorithms designed to detect target presence and localize the target to a specific location on the FPA of the EO sensor.