A 3D camera measures distance by comparing image blur, creating depth maps while using less power for wearables, robots, drones and assistive devices.
Northwestern University engineers have developed SpiderCam, a 3D camera that generates real-time depth maps while consuming just 624 milliwatts of power. Inspired by the vision system of jumping spiders, the camera estimates distance by comparing blur differences between two images captured with slightly different focus settings.
Designed for battery-powered applications, SpiderCam could support wearables, assistive devices, robots and drones that require depth sensing without the power demands of conventional 3D cameras. The system produces depth maps at 32.5 frames per second while operating below one watt of power.
The technology will be presented at the Computer Vision Foundation’s Conference on Computer Vision and Pattern Recognition (CVPR) in Denver. The research is also available on the arXiv preprint server.
Unlike traditional 3D cameras that rely on multiple viewpoints or projected light to calculate depth, SpiderCam uses a passive imaging approach. It simultaneously captures two images of the same scene with different focus settings and analyzes changes in image sharpness to determine distance.
The concept comes from jumping spiders, which have multiple retinal layers in each eye. Each layer receives an image focused at a different distance, allowing the spider to estimate depth by comparing levels of sharpness.
To process depth information efficiently, the researchers implemented a custom algorithm on a field-programmable gate array (FPGA). The chip analyzes edge and texture sharpness differences between the two images and converts them into depth measurements in real time.
According to the research team, SpiderCam is the first passive FPGA-based 3D camera system to operate below one watt of power consumption.
The researchers plan to improve the optics, widen the field of view and integrate the technology into wearable devices and small robots. They are also exploring a custom chip design that could further reduce power consumption for applications where access to power is limited.
