Pedestrian safety pilot project wins award

Devising a system that detects a pedestrian crossing in the wrong place in the roadway and flashing a warning to drivers about the pedestrian seems like a straightforward solution to pull off, in part because systems for detecting wildlife in the roadway are already in place.

But it turns out that creating a warning system to detect humans in unexpected areas isn’t easy because no such system currently exists for roadways. That didn’t stop Lakewood’s Transportation Engineering Division from taking up the challenge as part of the West Colfax Transportation Safety Project. This work is particularly important because many of the fatalities and injuries on West Colfax have resulted from pedestrians crossing mid-block late at night where a motorist would not expect a pedestrian to cross.

The division started testing a video-based system along with two Light Detection and Ranging (LIDAR) devices to evaluate their effectiveness in this task. The LIDAR systems use lasers to measure and detect objects. Then Sony approached Lakewood with a product the company was developing to add to the testing, and as a result of the field evaluation, the solution utilizing Sony's AITRIOS achieved the best accuracy.

Lakewood and its partners conducted a pilot project to see if the system would accurately detect an object, such as a pedestrian, entering the roadway and then alert drivers in the area. It paid special attention to how adverse weather would impact the performance of the technology. If successful, the system would act similarly to certain wildlife crossings, which can detect wildlife in the vicinity and provide a visual signal to drivers to be extra alert when entering a stretch of road.

Sony’s AITRIOS-enabled and Leopard Imaging/AglaiaSense were able to train their AI-powered detection tool to almost 100 percent accuracy to detect a person at least three feet into West Colfax for over 200 feet of distance from the camera. The project also performed both controlled and uncontrolled testing. Controlled testing set scenarios at different distances and speeds, while the uncontrolled testing monitored the normal rhythms of the roadway for several weeks.

The product from Sony and Leopard Imaging/AglaiaSense recently received the EDGE AI Blueprint Gold Medal for this work (watch award’s video) for groundbreaking pedestrian detection technology. The EDGE AI Foundation is a global nonprofit that seeks to bring together researchers, businesses and policymakers to tackle big challenges.

“The Gold Blueprint Award was awarded because it embodies the spirit of what the EDGE AI Foundation stands for: bold collaboration between public and private sectors to solve pressing real-world challenges,” said EDGE AI Foundation Chief Financial Officer Pete Bernard. “This initiative tackles a critical and universal issue — road safety — by harnessing the power of EDGE AI in a scalable, actionable way. By openly sharing their approach through our Foundation’s platform, Lakewood is not only advancing innovation, but also setting a precedent for cities nationwide. It’s a blueprint for impact — and a catalyst for change.”

Lakewood Transportation Engineer Mike Whiteaker explained the challenges that the project faced.

“With wildlife, you just detect anything entering the roadway at that location (usually a gap in a fence). We wanted to detect pedestrians mid-block without detecting bicycles, motorcycles or cars entering from driveways,” he said. “We also have to keep detecting the pedestrian even if cars are driving around them.”

The system also had to account for numerous variables: pedestrians jaywalking at different distances from the intersection and at the far end of the detection zone; pedestrians moving through the detection zone at varying speeds or crossing the road and pausing at the median for at least 30 seconds; and pedestrians walking parallel to the detection zone.

The project resulted in Lakewood now having a preferred detector that the Transportation Engineering Division knows can work. The division is looking at locations where the detection system could be installed, which could happen sooner elsewhere in the city than on West Colfax Avenue because the construction for the West Colfax Transportation Safety Project is just getting under construction and will continue during the next two years. Additionally, evolving technology in the coming years could also help improve the detection system.