Penn State builds photomemristor to improve robotic vision

According to Interesting Engineering, researchers at Penn State have developed a new optical sensor component called a photomemristor that mimics aspects of the human eye. The article reports the device adapts to mixed, changing light from bright to dark in seconds and measures about half a millimeter across. Interesting Engineering also reports the component can convert light energy into electrical current to power advanced optical systems. The piece includes a direct quote from Larry Cheng, James L. Henderson Jr. Memorial Associate Professor of Engineering Science and Mechanics at Penn State: "Self-driving cars are exposed to a mixture of light levels in use, imagine the contrast of the dark sky with the bright headlights of other cars when driving at night."

Photomemristors combine light sensitivity with memory-like electrical behaviour, which in academic literature is proposed to enable per-pixel adaptation and temporal filtering at the sensor level. Companies and research groups exploring neuromorphic or retina-inspired sensors typically seek lower latency and improved dynamic range compared with conventional image sensors; industry work in this area often targets robustness to rapid illumination changes without relying solely on downstream neural-net postprocessing.

For practitioners, integrating sensor-level adaptation reduces the burden on perception stacks that must currently compensate for blown highlights or clipped shadows in software. Observed patterns in similar sensor research show that hardware changes often require co-design of calibration pipelines and retraining of perception models to fully realise benefits.

Indicators to follow include peer-reviewed publication or technical report from the Penn State team, open datasets or benchmark results showing improved detection under mixed-light scenarios, and any partner demonstrations with robotics or automotive platforms.