Toshiba shrinks solid-state lidar for transport infrastructure

A key improvement is within the silicon photo-multiplier (SiPM) light receiving integrated circuits which have light-receiving cells controlled by transistors.

Toshiba shrinks solid-state lidar for transport infrastructure

The new chips have smaller Transistor modules, replace buffer layers protecting the transistors with newly-developed insulating trenches between the transistors and the light-receiving cells (diagram right).

To recover sensitivity lost in using smaller transistors, a high-withstand voltage section has been added to raise the voltage input to the light-receivers.

“We have developed technologies essential for a compact, high-resolution, long-range [200m] solid-state lidar solution that is robust and simple to implement,” according to Toshiba scientist Akihide Sai. “Demand for such a solution is anticipated in both the autonomous driving and transportation infrastructure monitoring applications.”

Toshiba shrinks solid-state lidar for transport infrastructure

Overall the silicon photo-multipier is 75% and 50% more sensitive than Toshiba’s July 2020 demonstrator, With more ICs side-by-side, resolution is up 4x to 1200 x 80.

With improved component packaging, the overall size of the lidar projector and receiver is down to 350cm3 (left).

Working towards all-weather outdoor use, a temperature compensation mechanism has been added by automatically adjusting the voltage applied to the light-receiving cells.

Toshiba shrinks solid-state lidar for transport infrastructure

In a demonstration (right) obstacles were detected outdoors on a sunny day. The lidar identified a target (a cardboard box) and measured its rage at 50.98m (figure bottom left). Max detection range was ~300m. Test was at 1feame/s with a fixed angle.

Applications are foreseen at the road side warning of subsidence, landslides, snow cover or objects in the carriageway. “Current methods for monitoring transportation infrastructure rely on cameras, but their performance is degraded by low light levels and adverse weather conditions,” said the company.

There are two detection videos:

Three obstacles at night (1frame/s, fixed angle, ~200m detection)

Sunny day detection of a car parked at 20m and a building at 300m (20frame/s)